{"id":5704,"date":"2024-03-13T09:28:31","date_gmt":"2024-03-13T08:28:31","guid":{"rendered":"https:\/\/www.a2d.ai\/deciding-in-the-presence-of-imprecision\/"},"modified":"2024-03-13T09:28:31","modified_gmt":"2024-03-13T08:28:31","slug":"deciding-in-the-presence-of-imprecision","status":"publish","type":"post","link":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/","title":{"rendered":"Deciding in the presence of imprecision"},"content":{"rendered":"<style>\nh1 {color: black; font-family: \"Times New Roman\", Times, serif;}<br \/>\nh2 {color: black; font-family: \"Times New Roman\", Times, serif;}<br \/>\nh3 {color: black; font-family: \"Times New Roman\", Times, serif;}<br \/>\nh4 {color: black; font-family: \"Times New Roman\", Times, serif;}<br \/>\na {color: black; font-weight: normal; font-family: \"Times New Roman\", Times, serif; text-align: justify;}<br \/>\np {color: black; font-weight: normal; font-family: \"Times New Roman\", Times, serif; text-align: justify;}<br \/>\ntd {color: black; font-weight: normal; font-family: \"Times New Roman\", Times, serif;}<br \/>\nli {color: black; font-weight: normal; font-family: \"Times New Roman\", Times, serif; text-align: justify;}<br \/>\ncenter {color: black; font-weight: normal; font-family: \"Times New Roman\", Times, serif;}<br \/>\ntable {border-collapse: collapse;}<br \/>\ntd, th {color: black; font-weight: normal; font-family: \"Times New Roman\", Times, serif; text-align: center; border: 1px solid #dddddd;}<br \/>\nfigcaption {color: black; font-weight: normal; font-size: 80%; font-family: \"Times New Roman\", Times, serif; text-align: center;}<br \/>\ncaption {color: black; font-weight: normal; font-size: 80%; font-family: \"Times New Roman\", Times, serif; text-align: center;}<br \/>\n.ql-right-eqno {color: black; font-weight: normal; font-size: 80% !important; font-family: \"Times New Roman\", Times, serif; width: 4em !important}<br \/>\n<\/style>\n<div class=\"article\">\n<h1>How to decide in the presence of imprecision?<\/h1>\n<p style=\"text-align: left;\"><strong>   S. Sochacki, PhD in Computer Science <\/strong><\/p>\n<p style=\"text-align: left;\"><em><strong>   A2D, ZA Les Minimes, 1 rue de la Trinquette, 17000 La Rochelle, FRANCE<\/strong><\/em><\/p>\n<h2>1 Introduction<\/h2>\n<p>Imprecision\/uncertainty\/conflict<br \/>\nAI Act, explainability<\/p>\n<p>How do you decide in IT? What does an automatic system base its decisions on? Very often, this kind of system is based on a principle of decision rules (often probabilistic), or on a pignistic approach (a level entirely dedicated to decision-making and clearly separated from data modeling).  <\/p>\n<p>Although a very large proportion of image processing applications for the general public are geared towards the enhancement, restoration or processing of photographs, in an industrial context, the target applications are image-based decision-making. While the classes of information used for this purpose are well known (texture, shape, color), they are not all of the same interest in decision support. Various forms of learning have therefore been developed to identify the best possible combination of information. Dozens of solutions exist to answer this question, and some even combine several of these approaches to optimize decision-making as much as possible.   <\/p>\n<p>However, while not all information classes, and more specifically all information extraction operators (Zernike moments, D&#8217;Haralick attributes, to name but two), are of equal interest for decision-making purposes, the information extracted is judged to be of equal quality. And yet, image processing comes in part from signal processing, and thus from the world of physics and, more specifically, metrology. So we&#8217;re a long way from a binary world, where information is true or false, but certain as to its assignment. In metrology, measurement only exists in relation to a criterion of precision or tolerance. This criterion quantifies the range of values within which the measurement lies with a good degree of certainty. This point of view, typical of needle-based metrology, is often forgotten in the transition to digital technology. And yet, every scientist remains sceptical when announcing a result with more than 3 digits behind the decimal point (when this precision has not been justified beforehand).      <\/p>\n<p>A2D is committed to developing applications using decision-making algorithms to meet these new needs, while respecting the constraints linked to trusted AI (AI Act <a href=\"#footnote-1\"><img decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-262bcc0158d697c58f339ad8fa642a53_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#94;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"9\" width=\"5\" style=\"vertical-align: 6px;\"\/><\/a>, ISO\/IEC 42001 <a href=\"#footnote-2\"><img decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9129dfc8cd44dea5a511583766cf1b07_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#94;&#50;\" title=\"Rendered by QuickLaTeX.com\" height=\"9\" width=\"7\" style=\"vertical-align: 6px;\"\/><\/a>) and energy sobriety. In practice, we are designing data processing and analysis applications to understand outdoor scenes, digitized with a view to detecting precise areas of interest, depending on business uses. <\/p>\n<p>We also endeavor to provide, at all times during the processing of information, an explanation of the results obtained and the choices proposed by the machine.<\/p>\n<h2>2 Probabilistic approach<\/h2>\n<p>Since our aim is to propose a new extension to decision-making tools, we felt it was important to stay within the framework of mathematically-based tools, as opposed to rule-based tools. As probabilities are at the heart of these approaches, we have chosen to start this study from this formalism. <\/p>\n<p>Classically, we call probability the evaluation of the likelihood of an event. It is represented by a real number between 0 and 1, indicating the degree of risk (or chance) that the event will occur. In the typical example of a coin toss, the event &#8220;tails side up&#8221; has a probability of <img decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-872abde626b8cdf8e983a3345ee98925_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#49;&#47;&#50;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"25\" style=\"vertical-align: -5px;\"\/>, which means that after tossing the coin a very large number of times, the frequency of occurrence of &#8220;tails side up&#8221; will be <img decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-872abde626b8cdf8e983a3345ee98925_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#49;&#47;&#50;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"25\" style=\"vertical-align: -5px;\"\/>.  <\/p>\n<p>In probability, an event can be just about anything you can imagine, whether it happens or not. It could be that the weather will be fine tomorrow, or that a roll of a die will produce the number 6. The only constraint is to be able to verify the event; it&#8217;s perfectly possible to check whether the weather will be fine tomorrow (an event that we measure personally, the following day, according to our tastes in terms of sunshine, temperature, etc.), or whether we manage to obtain the number 6 with a die.  <\/p>\n<p>We&#8217;ll say that an event with a probability of 0 is impossible, while an event with a probability of 1 is certain.<\/p>\n<p>Note that probability is a tool for predicting real-world events, not for explaining them. It&#8217;s about studying sets by measuring them. <\/p>\n<h3>2.1 Definitions<\/h3>\n<p>An experiment is said to be random when its outcome cannot be predicted. Let&#8217;s define <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fb1e11e13c33c9d9ad92194939f45f74_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#69;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"10\" style=\"vertical-align: 0px;\"\/>, a <strong>random experiment<\/strong> whose result <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-707fcec15e450815730425a6607a1858_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"11\" style=\"vertical-align: 0px;\"\/> is an element of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>, the set of all possible outcomes, called the <strong>universe of possibilities<\/strong> or <strong>frame of reference<\/strong>. <\/p>\n<p>Let <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-011a752786abb3a1fd4acd86bbf9b2ed_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#80;&#125;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"40\" style=\"vertical-align: -5px;\"\/> be the<strong>set of parts<\/strong> of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>. An event is then a logical proposition linked to an experiment, relative to its result. <\/p>\n<p>Let <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2da3234c3fdae205092efb71f0fb7572_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"15\" style=\"vertical-align: -1px;\"\/> be the set of events and subset of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5bed2094d6826d2cb5b8f63cef61b30e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#80;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: -1px;\"\/>. For all <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fa9557eabf8f805a8cb71793deb0a470_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#92;&#105;&#110;&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"50\" style=\"vertical-align: -1px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cbc4525d00d543bf8a600f7592246cc_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;&#92;&#105;&#110;&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"51\" style=\"vertical-align: -1px;\"\/> : <\/p>\n<ul>\n<li>A \\cup B<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0fa9ccba9d3b3af867358fea1503a610_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#100;&#101;&#115;&#105;&#103;&#110;&#97;&#116;&#101;&#115;&#32;&#116;&#104;&#101;&#32;&#114;&#101;&#97;&#108;&#105;&#122;&#97;&#116;&#105;&#111;&#110;&#32;&#111;&#102;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"216\" style=\"vertical-align: -4px;\"\/>A<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3e71f7c1fea563f41b81c6c9e32f6ab5_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#60;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#111;&#114;&#60;&#47;&#115;&#116;&#114;&#111;&#110;&#103;&#62;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"212\" style=\"vertical-align: -5px;\"\/>B<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3f792ed211878ce638f12a4d799bceb1_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#60;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#46;&#60;&#47;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#32;&#60;&#47;&#108;&#105;&#62; &#32;\t&#60;&#108;&#105;&#62;&#65;&#32;&#92;&#99;&#97;&#112;&#32;&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"360\" style=\"vertical-align: -5px;\"\/> denotes the realization of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> <strong>and<\/strong> <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> <strong>.<\/strong> <\/li>\n<li><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5bffc9df1006695edaf7c8e3a9ab7db5_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#98;&#97;&#114;&#123;&#65;&#125;&#61;&#92;&#79;&#109;&#101;&#103;&#97;&#92;&#98;&#97;&#99;&#107;&#115;&#108;&#97;&#115;&#104;&#32;&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"72\" style=\"vertical-align: -5px;\"\/> denotes the opposite of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/>.<\/li>\n<li><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> is the certain event<\/li>\n<li><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f46843de9b393af5b115a2a737d95fe0_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"8\" style=\"vertical-align: -1px;\"\/> is the impossible event<\/li>\n<\/ul>\n<p>When the reference frame <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> is finite, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2da3234c3fdae205092efb71f0fb7572_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"15\" style=\"vertical-align: -1px;\"\/> designates all the parts of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>, usually noted <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2b56ee3ef0153878226308be22d08790_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#50;&#94;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"19\" style=\"vertical-align: 0px;\"\/>. When the reference frame is <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a6e421454947c585b8fb5ae10299f873_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#98;&#98;&#123;&#82;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"13\" style=\"vertical-align: 0px;\"\/> (or an interval of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a6e421454947c585b8fb5ae10299f873_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#98;&#98;&#123;&#82;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"13\" style=\"vertical-align: 0px;\"\/>), the notion of <strong>tribe<\/strong> makes it possible to define <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2da3234c3fdae205092efb71f0fb7572_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"15\" style=\"vertical-align: -1px;\"\/> : <\/p>\n<ul>\n<li>\\mathcal{A}\\subseteq\\mathcal{P}(\\Omega)<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-474513f4003a139ff64090f0a86c1dfc_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#60;&#47;&#108;&#105;&#62; &#32;\t&#60;&#108;&#105;&#62;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"105\" style=\"vertical-align: -5px;\"\/>\\emptyset\\in\\mathcal{A}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-474513f4003a139ff64090f0a86c1dfc_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#60;&#47;&#108;&#105;&#62; &#32;\t&#60;&#108;&#105;&#62;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"105\" style=\"vertical-align: -5px;\"\/>\\forall\\{A_1,\\ldots,A_n\\}\\in\\mathcal{A}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-7bad522a2e34d09d63f372cc98ebc115_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#44;\" title=\"Rendered by QuickLaTeX.com\" height=\"6\" width=\"3\" style=\"vertical-align: -4px;\"\/>\\bigcup_iA_i\\in\\mathcal{A}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1d69d8b8af4015fa9e83e9b10ca96a88_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;&#110;&#100;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"30\" style=\"vertical-align: 0px;\"\/>\\bigcap_iA_i\\in\\mathcal{A}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-474513f4003a139ff64090f0a86c1dfc_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#60;&#47;&#108;&#105;&#62; &#32;\t&#60;&#108;&#105;&#62;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"105\" style=\"vertical-align: -5px;\"\/>\\bar{A}\\in\\mathcal{A}\\forall A\\in\\mathcal{A}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-63ccc96a9f45dc34f46fac047e0b55d9_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#60;&#47;&#108;&#105;&#62; &#60;&#47;&#117;&#108;&#62; &#65;&#32;&#60;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#109;&#101;&#97;&#115;&#117;&#114;&#101;&#60;&#47;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#32;&#111;&#110;&#32;&#97;&#32;&#109;&#101;&#97;&#115;&#117;&#114;&#97;&#98;&#108;&#101;&#32;&#115;&#112;&#97;&#99;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"40\" width=\"582\" style=\"vertical-align: -4px;\"\/>(\\Omega,\\mathcal{A})<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2235d662d3a36bf160167bfd78989c44_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;&#115;&#32;&#97;&#32;&#102;&#117;&#110;&#99;&#116;&#105;&#111;&#110;&#32;&#111;&#102;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"114\" style=\"vertical-align: -4px;\"\/>\\mathcal{A}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8c79d99d2ebdcf4428f34aa448f3ec56_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;&#110;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"17\" style=\"vertical-align: 0px;\"\/>[0,1]<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-68893b0dfebc9df752c82761f9ef3a94_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#115;&#117;&#99;&#104;&#32;&#116;&#104;&#97;&#116;&#32;&#58; &#60;&#117;&#108;&#62; &#32;\t&#60;&#108;&#105;&#62;&#80;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#61;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"244\" style=\"vertical-align: -5px;\"\/>1<\/li>\n<li>For all <strong>incompatible<\/strong> elements <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> (i.e. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2d5cec5bcdec0edbcc3e7728b732bb52_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#92;&#99;&#97;&#112;&#32;&#66;&#61;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;\" title=\"Rendered by QuickLaTeX.com\" height=\"17\" width=\"141\" style=\"vertical-align: -4px;\"\/>), <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3f2255a8145861c0df27097b520d30d4_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#92;&#99;&#117;&#112;&#32;&#66;&#41;&#61;&#80;&#40;&#65;&#41;&#43;&#80;&#40;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"202\" style=\"vertical-align: -5px;\"\/>.<\/li>\n<\/ul>\n<p>The number <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fcfc9b470d45ef45ab04f96571e6b70e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"40\" style=\"vertical-align: -5px;\"\/> quantifies how likely the event <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-28e1f1a318e2d41bc60fab4658efb076_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"49\" style=\"vertical-align: -3px;\"\/> is.<\/p>\n<p>In the case of a throw of the dice, we can state that <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3e87fb7a1b776b69d4cc76731f50aad0_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;&#61;&#92;&#123;&#49;&#44;&#50;&#44;&#51;&#44;&#52;&#44;&#53;&#44;&#54;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"146\" style=\"vertical-align: -5px;\"\/> represents all 6 faces of the dice. Thus, the probability of obtaining a number between 1 and 6 is certain, and therefore equal to 1. We&#8217;ll write <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fde01baeff30c799f3bb04c0c1491f18_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#49;&#92;&#99;&#117;&#112;&#32;&#50;&#92;&#99;&#117;&#112;&#32;&#51;&#92;&#99;&#117;&#112;&#32;&#52;&#92;&#99;&#117;&#112;&#32;&#53;&#92;&#99;&#117;&#112;&#32;&#54;&#41;&#61;&#80;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#61;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"275\" style=\"vertical-align: -5px;\"\/>.  <\/p>\n<p>Similarly, the probability of obtaining 1 (or 2, or 3, or 4 etc.) is 1\/6. So, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a89cdb08c6f467c7fbff7fc06a720ab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#49;&#41;&#61;&#80;&#40;&#50;&#41;&#61;&#46;&#46;&#46;&#61;&#80;&#40;&#54;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#54;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"22\" width=\"228\" style=\"vertical-align: -6px;\"\/>. <\/p>\n<p>Related to <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-650eb7688af6737ac325425b5c9a5982_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>, in the case where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> is finite, a probability distribution is defined as a function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3bf85f1087e9fbed3a319341134ac1a2_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"10\" style=\"vertical-align: -4px;\"\/> of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> in <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b6d943ab489c05a3dbd5ea29087a48_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#91;&#48;&#44;&#49;&#93;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"32\" style=\"vertical-align: -5px;\"\/>, such that:<br \/>\n<a name=\"id3648582144\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 39px;\"><span class=\"ql-right-eqno\"> (2.1.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5174fee6c5a520b29d3175fd8f33e4ac_l3.png\" height=\"39\" width=\"248\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#80;&#40;&#65;&#41;&#61;&#92;&#115;&#117;&#109;&#95;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#92;&#105;&#110;&#32;&#65;&#125;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#41;&#32;&#92;&#59;&#92;&#59;&#92;&#59;&#32;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#65;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#92;&#79;&#109;&#101;&#103;&#97; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>with the normalization condition:<br \/>\n<a name=\"id1914577839\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 41px;\"><span class=\"ql-right-eqno\"> (2.1.2) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1de9fec32938b2eb411e3057713fad5e_l3.png\" height=\"41\" width=\"131\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#115;&#117;&#109;&#95;&#123;&#111;&#109;&#101;&#103;&#97;&#105;&#110;&#92;&#79;&#109;&#101;&#103;&#97;&#125;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#41;&#61;&#49; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>Note that <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d60314f5c11aa273b265a5844e77b660_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#41;&#43;&#80;&#40;&#92;&#98;&#97;&#114;&#123;&#65;&#125;&#41;&#61;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"135\" style=\"vertical-align: -5px;\"\/>.<\/p>\n<h3>2.2 Bayes&#8217; rule<\/h3>\n<p>The introduction of a statistical model leads to a decision theory directly applicable, in principle, to the problem of classification. This application is essentially based on Bayes&#8217; rule, which makes it possible to evaluate probabilities a posteriori (i.e. after the actual observation of certain quantities) knowing the conditional probability distributions a priori (i.e. independent of any constraints on the observed variables). The probabilities required to make a decision according to Bayes&#8217; rule are provided by one (or more) statistical classifiers, which indicate the probability of a given individual belonging to each class of the problem.  <\/p>\n<p>In probability theory, Bayes&#8217; theorem states conditional probabilities: given two events <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>, Bayes&#8217; theorem allows us to determine the probability of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> knowing <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>, if we know the probabilities: of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/>, of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> and of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> knowing A. To arrive at Bayes&#8217; theorem, we need to start from one of the definitions of conditional probability: <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5a0acd7901cb84fa07a90d97bd220db5_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#124;&#66;&#41;&#80;&#40;&#66;&#41;&#61;&#80;&#40;&#65;&#32;&#92;&#99;&#97;&#112;&#32;&#66;&#41;&#61;&#80;&#40;&#66;&#124;&#65;&#41;&#80;&#40;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"324\" style=\"vertical-align: -5px;\"\/>, noting <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2c9dc7596f3e6abfa057ba5586d93883_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#32;&#92;&#99;&#97;&#112;&#32;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"74\" style=\"vertical-align: -5px;\"\/> the probability that <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> will both occur. Dividing on both sides by <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d175d9720aaed3998c4f33fe0f496f7a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"41\" style=\"vertical-align: -5px;\"\/>, we obtain <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f474a0c1a66f456e9357a439d54f20ee_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#124;&#66;&#41;&#61;&#80;&#40;&#66;&#124;&#65;&#41;&#80;&#40;&#65;&#41;&#47;&#80;&#40;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"235\" style=\"vertical-align: -5px;\"\/>, i.e. Bayes&#8217; theorem. Each term in Bayes&#8217; theorem has a common name. The term <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fcfc9b470d45ef45ab04f96571e6b70e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"40\" style=\"vertical-align: -5px;\"\/> is the <strong>a priori probability<\/strong> of A. It is &#8220;prior&#8221; in the sense that it precedes any information about <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fcfc9b470d45ef45ab04f96571e6b70e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"40\" style=\"vertical-align: -5px;\"\/> is also called the <strong>marginal probability<\/strong> of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/>. The term <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-cc5c85f8c782fa912dffae763b3a122f_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#65;&#124;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"59\" style=\"vertical-align: -5px;\"\/> is called the <strong>a posteriori probability<\/strong> of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> knowing <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> (or of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> under condition <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>). It is &#8220;posterior&#8221;, in the sense that it depends directly on <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>. The term <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f959961ae293baa55b9260b0e50f790c_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#66;&#124;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"59\" style=\"vertical-align: -5px;\"\/>, for a known <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>, is called the <strong>likelihood function<\/strong> of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/>. Similarly, the term <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d175d9720aaed3998c4f33fe0f496f7a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"41\" style=\"vertical-align: -5px;\"\/> is called the marginal or a priori probability of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>.      <\/p>\n<p>But, as previously stated, the probability values used by Bayes&#8217; rule are provided by statistical classifiers, or simply by estimating the probability distributions of each class.<\/p>\n<h3>2.3 Statistical decision-making<\/h3>\n<p>From the hypothesis about the probability distributions of the problem&#8217;s classes (and its validation tests) to the measurement of the distance to the center of each class, via various forms of model estimation, there are a huge number of methods in the field of statistics. Our aim is not to make an exhaustive study of them, but to compare theories (statistics, decision theory&#8230;) on the principle.<\/p>\n<p>Let&#8217;s just say that there are two main types of statistical approach: linear classifiers and quadratic classifiers. The former are so called because their approach consists of partitioning the attribute space in a simple way, based on straight lines, each straight line symbolizing the boundary between two classes, as illustrated by the boundary in green in figure 2.3.1. The second approach is more complex, as it attempts to estimate the shape of the boundaries between regions, thus obtaining the circular separation between classes.  <\/p>\n<figure id=\"attachment_3234\" style=\"width: 600px; margin-left: auto; margin-right: auto;\"><img loading=\"lazy\" decoding=\"async\" class=\"size-medium wp-image-3234\" src=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/exlinquad-300x225.png\" alt=\"exlinquad\" width=\"600\" height=\"450\" srcset=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/exlinquad-300x225.png 300w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/exlinquad.png 561w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/><figcaption style=\"text-align: center; margin-top: 0pt; padding-top: 0; margin-bottom: 10pt; padding-bottom: 0; line-height: 1; font-style: italic; color: #0e2841; font-size: 9pt;\">Figure 2.3.1 &#8211; Boundaries using a linear classifier (Bayes-Normal-1) and a quadratic classifier (Bayes-Normal-2)<\/figcaption><\/figure>\n<h4>2.3.1 Bayesian classifier [93][4][14]<\/h4>\n<p>Bayes decision theory is a statistical approach to the problem. It is based on the following assumptions: <\/p>\n<ul>\n<li>The decision problem is expressed in terms of probabilities<\/li>\n<li>The value of the a priori probabilities of drawing a class is known<\/li>\n<\/ul>\n<p>Let <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-707fcec15e450815730425a6607a1858_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"11\" style=\"vertical-align: 0px;\"\/> be a random variable representing the individual&#8217;s state (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-64c55e60b50932229832fc793b8675ae_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"17\" style=\"vertical-align: -6px;\"\/> if the individual belongs to class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-43c82d5bb00a7568d935a12e3bd969dd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"9\" style=\"vertical-align: -4px;\"\/>).<\/p>\n<p>Let <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-82f82a4590700fef69b481eb6e51d444_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;&#61;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#108;&#100;&#111;&#116;&#115;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#83;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"132\" style=\"vertical-align: -5px;\"\/> be the set of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-520cb534cd5b6bed768a61515b57cb7e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#83;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> possible states or classes of a problem.<\/p>\n<p>The set of <strong>a priori<\/strong> probabilities <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3459c7452b5d3cb703760cf4f2294117_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#41;&#44;&#32;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;&#41;&#44;&#32;&#92;&#108;&#100;&#111;&#116;&#115;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"119\" style=\"vertical-align: -5px;\"\/> is calculated from observations made on the system under study. In the absence of other information, we have no choice but to write <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1babc352b331ddc8ac0bc164358a6ccd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#40;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"16\" style=\"vertical-align: -5px;\"\/>individual <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fac4ce39409b0f32f27dd03535ff4371_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#61;&#32;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#105;&#41;&#61;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#105;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"104\" style=\"vertical-align: -5px;\"\/>. But if it is possible to obtain other information (measurements), then <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-119433038c963294db2d93adf82d09ec_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#40;&#120;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"56\" style=\"vertical-align: -6px;\"\/>, the probability density of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/> <strong>knowing<\/strong> that the individual is <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-64c55e60b50932229832fc793b8675ae_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"17\" style=\"vertical-align: -6px;\"\/>, or the probability density conditional on his belonging to <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-64c55e60b50932229832fc793b8675ae_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"17\" style=\"vertical-align: -6px;\"\/>, can be calculated.  <\/p>\n<p>p(\\omega_j)<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-caa89898cceb982bed2c9331375d51f6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;&#115;&#32;&#116;&#104;&#101;&#114;&#101;&#102;&#111;&#114;&#101;&#32;&#116;&#104;&#101;&#32;&#97;&#32;&#112;&#114;&#105;&#111;&#114;&#105;&#32;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#111;&#102;&#32;&#97;&#112;&#112;&#101;&#97;&#114;&#97;&#110;&#99;&#101;&#32;&#111;&#102;&#32;&#116;&#104;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"409\" style=\"vertical-align: -4px;\"\/>\\omega_j<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5f944eae7fea4cd925792b60df96491d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#99;&#108;&#97;&#115;&#115;&#32;&#97;&#110;&#100;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"69\" style=\"vertical-align: 0px;\"\/>p(\\omega_j|x)<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0a141d2285836e67650d6f5c6d8d8aed_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;&#115;&#32;&#116;&#104;&#101;&#32;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#116;&#104;&#97;&#116;&#32;&#116;&#104;&#101;&#32;&#115;&#104;&#97;&#112;&#101;&#32;&#99;&#104;&#97;&#114;&#97;&#99;&#116;&#101;&#114;&#105;&#122;&#101;&#100;&#32;&#98;&#121;&#32;&#116;&#104;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"375\" style=\"vertical-align: -4px;\"\/>x<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d51a4b381f8db7e3e4d82143273935a0_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#118;&#101;&#99;&#116;&#111;&#114;&#32;&#98;&#101;&#108;&#111;&#110;&#103;&#115;&#32;&#116;&#111;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"121\" style=\"vertical-align: -4px;\"\/>\\omega_j<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-17af2cf5031374d573dfd7809fa6ea65_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#46;  &#73;&#110;&#32;&#116;&#104;&#105;&#115;&#32;&#99;&#97;&#115;&#101;&#44;&#32;&#119;&#101;&#32;&#104;&#97;&#118;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"28\" width=\"160\" style=\"vertical-align: -4px;\"\/>x=\\{x_1 \\ldots x_d\\}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a15050225dc7431807627b593ec542dd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#44;&#32;&#116;&#104;&#101;&#32;&#114;&#97;&#110;&#100;&#111;&#109;&#32;&#118;&#101;&#99;&#116;&#111;&#114;&#32;&#40;&#109;&#101;&#97;&#115;&#117;&#114;&#101;&#109;&#101;&#110;&#116;&#115;&#32;&#111;&#114;&#32;&#115;&#105;&#116;&#117;&#97;&#116;&#105;&#111;&#110;&#41;&#32;&#111;&#110;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"383\" style=\"vertical-align: -5px;\"\/>{mathbb{R}}^d<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f07904efb01f148f96dfd3594107a036_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#46;&#32;&#84;&#104;&#101;&#32;&#112;&#114;&#111;&#98;&#108;&#101;&#109;&#32;&#116;&#104;&#101;&#110;&#32;&#98;&#101;&#99;&#111;&#109;&#101;&#115;&#32;&#58;&#32;&#32;&#32;&#60;&#101;&#109;&#62;&#105;&#115;&#32;&#97;&#110;&#32;&#105;&#110;&#100;&#105;&#118;&#105;&#100;&#117;&#97;&#108;&#32;&#97;&#110;&#100;&#32;&#105;&#116;&#115;&#32;&#109;&#101;&#97;&#115;&#117;&#114;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"508\" style=\"vertical-align: -4px;\"\/>x<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-735b8e584e12035ed468587fa73eb034_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#46;&#32;&#72;&#111;&#119;&#32;&#100;&#111;&#32;&#121;&#111;&#117;&#32;&#100;&#101;&#99;&#105;&#100;&#101;&#32;&#119;&#104;&#105;&#99;&#104;&#32;&#99;&#97;&#116;&#101;&#103;&#111;&#114;&#121;&#32;&#40;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"257\" style=\"vertical-align: -5px;\"\/>\\omega_j<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 247px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b31a8f2db2a9e6633c075c4011b241ef_l3.png\" height=\"247\" width=\"664\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#41;&#32;&#116;&#111;&#32;&#97;&#115;&#115;&#105;&#103;&#110;&#32;&#116;&#104;&#105;&#115;&#32;&#105;&#110;&#100;&#105;&#118;&#105;&#100;&#117;&#97;&#108;&#32;&#116;&#111;&#63;&#32;&#60;&#47;&#101;&#109;&#62; &#66;&#97;&#121;&#101;&#115;&#39;&#32;&#114;&#117;&#108;&#101;&#32;&#103;&#105;&#118;&#101;&#115;&#32;&#116;&#104;&#101;&#32;&#97;&#110;&#115;&#119;&#101;&#114;&#58; &#60;&#97;&#32;&#110;&#97;&#109;&#101;&#61;&#34;&#105;&#100;&#50;&#54;&#51;&#54;&#48;&#51;&#48;&#51;&#52;&#55;&#34;&#62;&#60;&#47;&#97;&#62;&#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#114;&#105;&#103;&#104;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#40;&#50;&#46;&#51;&#46;&#49;&#46;&#49;&#41;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#108;&#101;&#102;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#32;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#105;&#109;&#103;&#32;&#115;&#114;&#99;&#61;&#34;&#104;&#116;&#116;&#112;&#115;&#58;&#47;&#47;&#119;&#119;&#119;&#46;&#97;&#50;&#100;&#46;&#97;&#105;&#47;&#119;&#112;&#45;&#99;&#111;&#110;&#116;&#101;&#110;&#116;&#47;&#113;&#108;&#45;&#99;&#97;&#99;&#104;&#101;&#47;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#46;&#99;&#111;&#109;&#45;&#56;&#57;&#57;&#57;&#55;&#55;&#48;&#48;&#48;&#101;&#98;&#49;&#99;&#98;&#98;&#48;&#56;&#51;&#57;&#102;&#57;&#55;&#55;&#54;&#56;&#52;&#50;&#97;&#97;&#52;&#98;&#57;&#95;&#108;&#51;&#46;&#112;&#110;&#103;&#34;&#32;&#104;&#101;&#105;&#103;&#104;&#116;&#61;&#34;&#52;&#51;&#34;&#32;&#119;&#105;&#100;&#116;&#104;&#61;&#34;&#50;&#56;&#50;&#34;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#105;&#109;&#103;&#45;&#100;&#105;&#115;&#112;&#108;&#97;&#121;&#101;&#100;&#45;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#32;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#45;&#97;&#117;&#116;&#111;&#45;&#102;&#111;&#114;&#109;&#97;&#116;&#34;&#32;&#97;&#108;&#116;&#61;&#34;&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#92;&#112;&#114;&#111;&#116;&#101;&#99;&#116;&#32;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;&#124;&#120;&#41;&#32;&#61;&#32;&#92;&#102;&#114;&#97;&#99;&#123;&#112;&#40;&#120;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;&#41;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;&#41;&#125;&#123;&#112;&#40;&#120;&#41;&#125;&#32;&#92;&#44;&#92;&#44;&#92;&#44;&#32;&#44;&#32;&#92;&#44;&#92;&#44;&#92;&#44;&#32;&#106;&#92;&#105;&#110;&#92;&#123;&#49;&#44;&#83;&#92;&#125;&#32;&#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#34;&#32;&#116;&#105;&#116;&#108;&#101;&#61;&#34;&#82;&#101;&#110;&#100;&#101;&#114;&#101;&#100;&#32;&#98;&#121;&#32;&#81;&#117;&#105;&#99;&#107;&#76;&#97;&#84;&#101;&#88;&#46;&#99;&#111;&#109;&#34;&#47;&#62; &#119;&#105;&#116;&#104; &#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#114;&#105;&#103;&#104;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#40;&#50;&#46;&#51;&#46;&#49;&#46;&#50;&#41;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#108;&#101;&#102;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#32;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#105;&#109;&#103;&#32;&#115;&#114;&#99;&#61;&#34;&#104;&#116;&#116;&#112;&#115;&#58;&#47;&#47;&#119;&#119;&#119;&#46;&#97;&#50;&#100;&#46;&#97;&#105;&#47;&#119;&#112;&#45;&#99;&#111;&#110;&#116;&#101;&#110;&#116;&#47;&#113;&#108;&#45;&#99;&#97;&#99;&#104;&#101;&#47;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#46;&#99;&#111;&#109;&#45;&#51;&#102;&#99;&#101;&#57;&#52;&#99;&#100;&#53;&#48;&#55;&#101;&#102;&#54;&#97;&#101;&#101;&#99;&#48;&#54;&#53;&#49;&#57;&#51;&#54;&#48;&#51;&#102;&#48;&#55;&#101;&#102;&#95;&#108;&#51;&#46;&#112;&#110;&#103;&#34;&#32;&#104;&#101;&#105;&#103;&#104;&#116;&#61;&#34;&#53;&#53;&#34;&#32;&#119;&#105;&#100;&#116;&#104;&#61;&#34;&#49;&#56;&#50;&#34;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#105;&#109;&#103;&#45;&#100;&#105;&#115;&#112;&#108;&#97;&#121;&#101;&#100;&#45;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#32;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#45;&#97;&#117;&#116;&#111;&#45;&#102;&#111;&#114;&#109;&#97;&#116;&#34;&#32;&#97;&#108;&#116;&#61;&#34;&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#112;&#40;&#120;&#41;&#32;&#61;&#32;&#92;&#115;&#117;&#109;&#95;&#123;&#106;&#61;&#49;&#125;&#94;&#123;&#83;&#125;&#112;&#40;&#120;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;&#41;&#112;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#106;&#41;&#32;&#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#34;&#32;&#116;&#105;&#116;&#108;&#101;&#61;&#34;&#82;&#101;&#110;&#100;&#101;&#114;&#101;&#100;&#32;&#98;&#121;&#32;&#81;&#117;&#105;&#99;&#107;&#76;&#97;&#84;&#101;&#88;&#46;&#99;&#111;&#109;&#34;&#47;&#62; &#119;&#104;&#101;&#114;&#101;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>S<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-81d94f36cc9e108cff379454eb48aa55_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;&#115;&#32;&#116;&#104;&#101;&#32;&#110;&#117;&#109;&#98;&#101;&#114;&#32;&#111;&#102;&#32;&#100;&#105;&#102;&#102;&#101;&#114;&#101;&#110;&#116;&#32;&#99;&#108;&#97;&#115;&#115;&#101;&#115;&#32;&#97;&#110;&#100;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"283\" style=\"vertical-align: -4px;\"\/>p(x)<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e20c036c98ef19ebc822888e5053f0e8_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#116;&#104;&#101;&#32;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#100;&#105;&#115;&#116;&#114;&#105;&#98;&#117;&#116;&#105;&#111;&#110;&#32;&#102;&#111;&#114;&#32;&#97;&#108;&#108;&#32;&#105;&#110;&#100;&#105;&#118;&#105;&#100;&#117;&#97;&#108;&#115;&#32;&#40;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#100;&#101;&#110;&#115;&#105;&#116;&#121;&#32;&#111;&#102;&#32;&#118;&#101;&#99;&#116;&#111;&#114;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"558\" style=\"vertical-align: -5px;\"\/>x<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1f508edd54ef1731d268669edbdc4371_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#41;&#44;&#32;&#105;&#110;&#32;&#119;&#104;&#105;&#99;&#104;&#32;&#116;&#104;&#101;&#32;&#115;&#97;&#109;&#112;&#108;&#101;&#115;&#32;&#111;&#102;&#32;&#101;&#97;&#99;&#104;&#32;&#99;&#108;&#97;&#115;&#115;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"260\" style=\"vertical-align: -5px;\"\/>\\omega_j<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a72e9e1dec4ce941c0163d9d26f2ddf0_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;&#112;&#112;&#101;&#97;&#114;&#32;&#60;&#101;&#109;&#62;&#112;&#114;&#111;&#112;&#111;&#114;&#116;&#105;&#111;&#110;&#97;&#108;&#108;&#121;&#32;&#116;&#111;&#60;&#47;&#101;&#109;&#62;&#32;&#116;&#104;&#101;&#32;&#60;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#97;&#32;&#112;&#114;&#105;&#111;&#114;&#105;&#60;&#47;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#32;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#105;&#101;&#115;\" title=\"Rendered by QuickLaTeX.com\" height=\"41\" width=\"582\" style=\"vertical-align: -5px;\"\/>p(\\omega_j)<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9593beb88a5c44efaf4c4c279c043275_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#59;&#32;&#116;&#104;&#105;&#115;&#32;&#105;&#115;&#32;&#116;&#104;&#101;&#32;&#110;&#111;&#114;&#109;&#97;&#108;&#105;&#122;&#97;&#116;&#105;&#111;&#110;&#32;&#102;&#97;&#99;&#116;&#111;&#114;&#46;  &#84;&#104;&#101;&#32;&#100;&#101;&#99;&#105;&#115;&#105;&#111;&#110;&#32;&#114;&#117;&#108;&#101;&#32;&#105;&#115;&#32;&#116;&#104;&#101;&#110;&#32;&#97;&#115;&#32;&#102;&#111;&#108;&#108;&#111;&#119;&#115;&#58;  &#60;&#99;&#101;&#110;&#116;&#101;&#114;&#62;&#100;&#101;&#99;&#105;&#100;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"60\" width=\"275\" style=\"vertical-align: -4px;\"\/>\\omega_j<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-32bec29a021ef3f88862d4de60e67d57_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;&#102;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"16\" style=\"vertical-align: -4px;\"\/>p(\\omega_j|x)=max_{i=1,n}(p(\\omega_i|x))<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b3e94270ab953bb0409f8a3e7a4f6ae6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#84;&#104;&#117;&#115;&#60;&#47;&#99;&#101;&#110;&#116;&#101;&#114;&#62;&#44;&#32;&#105;&#116;&#32;&#98;&#101;&#99;&#111;&#109;&#101;&#115;&#32;&#112;&#111;&#115;&#115;&#105;&#98;&#108;&#101;&#32;&#116;&#111;&#32;&#101;&#110;&#117;&#109;&#101;&#114;&#97;&#116;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"388\" style=\"vertical-align: -5px;\"\/>A={alpha_1,\\ldots,\\alpha_k\\}<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8b768ffd7dbfe3b8cd38430bc412b0f8_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#116;&#104;&#101;&#32;&#115;&#101;&#116;&#32;&#111;&#102;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"66\" style=\"vertical-align: -4px;\"\/>a<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-58450a1f43669ed7d69c22b2e98b0907_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#111;&#115;&#115;&#105;&#98;&#108;&#101;&#32;&#100;&#101;&#99;&#105;&#115;&#105;&#111;&#110;&#115;&#46;  &#78;&#111;&#119;&#32;&#119;&#101;&#32;&#110;&#101;&#101;&#100;&#32;&#116;&#111;&#32;&#101;&#120;&#112;&#114;&#101;&#115;&#115;&#32;&#116;&#104;&#101;&#32;&#99;&#111;&#110;&#100;&#105;&#116;&#105;&#111;&#110;&#97;&#108;&#32;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#100;&#101;&#110;&#115;&#105;&#116;&#105;&#101;&#115;&#46;&#32;&#84;&#111;&#32;&#100;&#111;&#32;&#116;&#104;&#105;&#115;&#44;&#32;&#108;&#101;&#116;&#39;&#115;&#32;&#114;&#101;&#99;&#97;&#108;&#108;&#32;&#116;&#104;&#101;&#60;&#101;&#109;&#62;&#66;&#105;&#101;&#110;&#97;&#121;&#109;&eacute;&#45;&#84;&#99;&#104;&#101;&#98;&#105;&#99;&#104;&#101;&#102;&#102;&#32;&#105;&#110;&#101;&#113;&#117;&#97;&#108;&#105;&#116;&#121;&#60;&#47;&#101;&#109;&#62;&#58; &#76;&#101;&#116;\" title=\"Rendered by QuickLaTeX.com\" height=\"61\" width=\"583\" style=\"vertical-align: -5px;\"\/>X_1, X_2, &#8230;, X_n<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b4176d34d089911d0d03a3736bf78f2f_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#98;&#101;&#32;&#105;&#110;&#100;&#101;&#112;&#101;&#110;&#100;&#101;&#110;&#116;&#32;&#114;&#97;&#110;&#100;&#111;&#109;&#32;&#118;&#97;&#114;&#105;&#97;&#98;&#108;&#101;&#115;&#44;&#32;&#97;&#108;&#108;&#32;&#119;&#105;&#116;&#104;&#32;&#101;&#120;&#112;&#101;&#99;&#116;&#97;&#116;&#105;&#111;&#110;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"402\" style=\"vertical-align: -4px;\"\/>\\mu<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2e96a87861a07762abd7e485f77b9b5c_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;&#110;&#100;&#32;&#118;&#97;&#114;&#105;&#97;&#110;&#99;&#101;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"98\" style=\"vertical-align: 0px;\"\/>\\sigma^2<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0e319984023bcbf2a7b24614998ee226_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#46;&#32;&#76;&#101;&#116;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"30\" style=\"vertical-align: 0px;\"\/>\\bar{X}=\\frac  {1}{n}\\sum_{i=1}^{n}X_i<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f0f18f9fbffc9c4b7b3d83a5cb33f060_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#116;&#104;&#101;&#110;&#32;&#102;&#111;&#114;&#32;&#97;&#108;&#108;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"83\" style=\"vertical-align: -4px;\"\/>d&gt;0<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 286px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-734448e47980810d05fdf4b26ae032fa_l3.png\" height=\"286\" width=\"898\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#58; &#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#114;&#105;&#103;&#104;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#40;&#50;&#46;&#51;&#46;&#49;&#46;&#51;&#41;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#108;&#101;&#102;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#32;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#105;&#109;&#103;&#32;&#115;&#114;&#99;&#61;&#34;&#104;&#116;&#116;&#112;&#115;&#58;&#47;&#47;&#119;&#119;&#119;&#46;&#97;&#50;&#100;&#46;&#97;&#105;&#47;&#119;&#112;&#45;&#99;&#111;&#110;&#116;&#101;&#110;&#116;&#47;&#113;&#108;&#45;&#99;&#97;&#99;&#104;&#101;&#47;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#46;&#99;&#111;&#109;&#45;&#55;&#50;&#52;&#101;&#55;&#101;&#49;&#48;&#98;&#50;&#51;&#101;&#101;&#52;&#55;&#102;&#55;&#49;&#51;&#101;&#101;&#53;&#54;&#51;&#53;&#52;&#51;&#101;&#50;&#102;&#52;&#101;&#95;&#108;&#51;&#46;&#112;&#110;&#103;&#34;&#32;&#104;&#101;&#105;&#103;&#104;&#116;&#61;&#34;&#51;&#57;&#34;&#32;&#119;&#105;&#100;&#116;&#104;&#61;&#34;&#50;&#48;&#49;&#34;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#105;&#109;&#103;&#45;&#100;&#105;&#115;&#112;&#108;&#97;&#121;&#101;&#100;&#45;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#32;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#45;&#97;&#117;&#116;&#111;&#45;&#102;&#111;&#114;&#109;&#97;&#116;&#34;&#32;&#97;&#108;&#116;&#61;&#34;&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#80;&#40;&#124;&#92;&#98;&#97;&#114;&#123;&#88;&#125;&#45;&#92;&#109;&#117;&#124;&#92;&#103;&#101;&#113;&#32;&#100;&#41;&#92;&#108;&#101;&#113;&#32;&#92;&#102;&#114;&#97;&#99;&#123;&#115;&#105;&#103;&#109;&#97;&#94;&#50;&#125;&#123;&#110;&#100;&#94;&#50;&#125;&#32;&#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#34;&#32;&#116;&#105;&#116;&#108;&#101;&#61;&#34;&#82;&#101;&#110;&#100;&#101;&#114;&#101;&#100;&#32;&#98;&#121;&#32;&#81;&#117;&#105;&#99;&#107;&#76;&#97;&#84;&#101;&#88;&#46;&#99;&#111;&#109;&#34;&#47;&#62; &#60;&#101;&#109;&#62;&#84;&#104;&#101;&#32;&#112;&#114;&#111;&#98;&#97;&#98;&#105;&#108;&#105;&#116;&#121;&#32;&#111;&#102;&#32;&#97;&#32;&#114;&#97;&#110;&#100;&#111;&#109;&#32;&#118;&#97;&#114;&#105;&#97;&#98;&#108;&#101;&#32;&#100;&#101;&#118;&#105;&#97;&#116;&#105;&#110;&#103;&#32;&#98;&#121;&#32;&#109;&#111;&#114;&#101;&#32;&#116;&#104;&#97;&#110;&#60;&#47;&#101;&#109;&#62;&#32;&#100;&#32;&#60;&#101;&#109;&#62;&#102;&#114;&#111;&#109;&#32;&#105;&#116;&#115;&#32;&#109;&#101;&#97;&#110;&#32;&#118;&#97;&#108;&#117;&#101;&#32;&#105;&#115;&#32;&#108;&#111;&#119;&#101;&#114;&#32;&#116;&#104;&#101;&#32;&#115;&#109;&#97;&#108;&#108;&#101;&#114;&#32;&#105;&#116;&#115;&#32;&#118;&#97;&#114;&#105;&#97;&#110;&#99;&#101;&#32;&#97;&#110;&#100;&#32;&#116;&#104;&#101;&#60;&#47;&#101;&#109;&#62;&#32;&#60;&#101;&#109;&#62;&#108;&#97;&#114;&#103;&#101;&#114;&#60;&#47;&#101;&#109;&#62;&#32;&#100;&#32;&#60;&#101;&#109;&#62;&#105;&#115;&#46;&#60;&#47;&#101;&#109;&#62;&#32;&#73;&#110;&#32;&#112;&#97;&#114;&#116;&#105;&#99;&#117;&#108;&#97;&#114;&#60;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#40;&#76;&#97;&#119;&#32;&#111;&#102;&#32;&#76;&#97;&#114;&#103;&#101;&#32;&#78;&#117;&#109;&#98;&#101;&#114;&#115;&#60;&#47;&#115;&#116;&#114;&#111;&#110;&#103;&#62;&#41;&#32;&#91;&#55;&#93;&#58; &#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#114;&#105;&#103;&#104;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#40;&#50;&#46;&#51;&#46;&#49;&#46;&#52;&#41;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#108;&#101;&#102;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#32;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#105;&#109;&#103;&#32;&#115;&#114;&#99;&#61;&#34;&#104;&#116;&#116;&#112;&#115;&#58;&#47;&#47;&#119;&#119;&#119;&#46;&#97;&#50;&#100;&#46;&#97;&#105;&#47;&#119;&#112;&#45;&#99;&#111;&#110;&#116;&#101;&#110;&#116;&#47;&#113;&#108;&#45;&#99;&#97;&#99;&#104;&#101;&#47;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#46;&#99;&#111;&#109;&#45;&#97;&#54;&#102;&#48;&#97;&#49;&#56;&#97;&#49;&#52;&#48;&#56;&#56;&#49;&#54;&#100;&#49;&#100;&#54;&#50;&#48;&#100;&#53;&#55;&#51;&#99;&#57;&#97;&#102;&#51;&#100;&#100;&#95;&#108;&#51;&#46;&#112;&#110;&#103;&#34;&#32;&#104;&#101;&#105;&#103;&#104;&#116;&#61;&#34;&#50;&#49;&#34;&#32;&#119;&#105;&#100;&#116;&#104;&#61;&#34;&#50;&#57;&#53;&#34;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#105;&#109;&#103;&#45;&#100;&#105;&#115;&#112;&#108;&#97;&#121;&#101;&#100;&#45;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#32;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#45;&#97;&#117;&#116;&#111;&#45;&#102;&#111;&#114;&#109;&#97;&#116;&#34;&#32;&#97;&#108;&#116;&#61;&#34;&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#80;&#40;&#124;&#92;&#98;&#97;&#114;&#123;&#88;&#125;&#45;&#92;&#109;&#117;&#124;&#92;&#103;&#101;&#113;&#32;&#100;&#41;&#92;&#108;&#111;&#110;&#103;&#114;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#32;&#48;&#32;&#92;&#44;&#92;&#44;&#92;&#44;&#32;&#119;&#104;&#101;&#110;&#32;&#92;&#44;&#92;&#44;&#92;&#44;&#32;&#110;&#92;&#114;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#32;&#43;&#92;&#105;&#110;&#102;&#116;&#121;&#32;&#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#34;&#32;&#116;&#105;&#116;&#108;&#101;&#61;&#34;&#82;&#101;&#110;&#100;&#101;&#114;&#101;&#100;&#32;&#98;&#121;&#32;&#81;&#117;&#105;&#99;&#107;&#76;&#97;&#84;&#101;&#88;&#46;&#99;&#111;&#109;&#34;&#47;&#62; &#73;&#102;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>n\\geq 30<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-7a2a41e2978c36d487cb83fe037ce306_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#44;&#32;&#116;&#104;&#101;&#110;&#44;&#32;&#98;&#121;&#32;&#97;&#112;&#112;&#108;&#121;&#105;&#110;&#103;&#32;&#116;&#104;&#101;&#32;&#99;&#101;&#110;&#116;&#114;&#97;&#108;&#32;&#108;&#105;&#109;&#105;&#116;&#32;&#116;&#104;&#101;&#111;&#114;&#101;&#109;&#32;&#40;&#119;&#104;&#105;&#99;&#104;&#32;&#105;&#110;&#116;&#117;&#105;&#116;&#105;&#118;&#101;&#108;&#121;&#32;&#115;&#116;&#97;&#116;&#101;&#115;&#32;&#116;&#104;&#97;&#116;&#32;&#97;&#110;&#121;&#32;&#115;&#117;&#109;&#32;&#111;&#102;&#32;&#105;&#110;&#100;&#101;&#112;&#101;&#110;&#100;&#101;&#110;&#116;&#44;&#32;&#105;&#100;&#101;&#110;&#116;&#105;&#99;&#97;&#108;&#108;&#121;&#32;&#100;&#105;&#115;&#116;&#114;&#105;&#98;&#117;&#116;&#101;&#100;&#32;&#114;&#97;&#110;&#100;&#111;&#109;&#32;&#118;&#97;&#114;&#105;&#97;&#98;&#108;&#101;&#115;&#32;&#116;&#101;&#110;&#100;&#115;&#32;&#116;&#111;&#119;&#97;&#114;&#100;&#115;&#32;&#97;&#32;&#71;&#97;&#117;&#115;&#115;&#105;&#97;&#110;&#32;&#114;&#97;&#110;&#100;&#111;&#109;&#32;&#118;&#97;&#114;&#105;&#97;&#98;&#108;&#101;&#41;&#44;&#32;&#105;&#116;&#32;&#99;&#97;&#110;&#32;&#98;&#101;&#32;&#97;&#115;&#115;&#117;&#109;&#101;&#100;&#32;&#116;&#104;&#97;&#116;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"1521\" style=\"vertical-align: -5px;\"\/>X_1+X_2+&#8230;+X_n<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 259px;\"><span class=\"ql-right-eqno\"> (2.3.1.6) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e97e7ce957a44f58260e80538b3db0ac_l3.png\" height=\"259\" width=\"724\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#112;&#114;&#97;&#99;&#116;&#105;&#99;&#97;&#108;&#108;&#121;&#32;&#102;&#111;&#108;&#108;&#111;&#119;&#115;&#32;&#97;&#32;&#78;&#111;&#114;&#109;&#97;&#108;&#32;&#100;&#105;&#115;&#116;&#114;&#105;&#98;&#117;&#116;&#105;&#111;&#110;&#46;&#32;&#82;&#101;&#99;&#97;&#108;&#108;&#32;&#116;&#104;&#97;&#116;&#32;&#116;&#104;&#101;&#32;&#100;&#101;&#110;&#115;&#105;&#116;&#121;&#32;&#111;&#102;&#32;&#97;&#32;&#78;&#111;&#114;&#109;&#97;&#108;&#32;&#100;&#105;&#115;&#116;&#114;&#105;&#98;&#117;&#116;&#105;&#111;&#110;&#32;&#105;&#115;&#32;&#111;&#102;&#32;&#116;&#104;&#101;&#32;&#102;&#111;&#114;&#109;&#32;&#40;&#111;&#110;&#32;&#111;&#110;&#101;&#32;&#100;&#105;&#109;&#101;&#110;&#115;&#105;&#111;&#110;&#41;&#58; &#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#114;&#105;&#103;&#104;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#40;&#50;&#46;&#51;&#46;&#49;&#46;&#53;&#41;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#115;&#112;&#97;&#110;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#108;&#101;&#102;&#116;&#45;&#101;&#113;&#110;&#111;&#34;&#62;&#32;&#32;&#32;&#60;&#47;&#115;&#112;&#97;&#110;&#62;&#60;&#105;&#109;&#103;&#32;&#115;&#114;&#99;&#61;&#34;&#104;&#116;&#116;&#112;&#115;&#58;&#47;&#47;&#119;&#119;&#119;&#46;&#97;&#50;&#100;&#46;&#97;&#105;&#47;&#119;&#112;&#45;&#99;&#111;&#110;&#116;&#101;&#110;&#116;&#47;&#113;&#108;&#45;&#99;&#97;&#99;&#104;&#101;&#47;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#46;&#99;&#111;&#109;&#45;&#50;&#97;&#50;&#102;&#99;&#98;&#99;&#52;&#97;&#48;&#99;&#99;&#57;&#99;&#98;&#52;&#54;&#98;&#56;&#52;&#102;&#102;&#51;&#100;&#101;&#55;&#52;&#54;&#50;&#55;&#102;&#56;&#95;&#108;&#51;&#46;&#112;&#110;&#103;&#34;&#32;&#104;&#101;&#105;&#103;&#104;&#116;&#61;&#34;&#49;&#54;&#48;&#34;&#32;&#119;&#105;&#100;&#116;&#104;&#61;&#34;&#53;&#56;&#52;&#34;&#32;&#99;&#108;&#97;&#115;&#115;&#61;&#34;&#113;&#108;&#45;&#105;&#109;&#103;&#45;&#100;&#105;&#115;&#112;&#108;&#97;&#121;&#101;&#100;&#45;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#32;&#113;&#117;&#105;&#99;&#107;&#108;&#97;&#116;&#101;&#120;&#45;&#97;&#117;&#116;&#111;&#45;&#102;&#111;&#114;&#109;&#97;&#116;&#34;&#32;&#97;&#108;&#116;&#61;&#34;&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#120;&#32;&#92;&#115;&#105;&#109;&#32;&#78;&#40;&#92;&#109;&#117;&#44;&#32;&#92;&#115;&#105;&#103;&#109;&#97;&#94;&#50;&#41;&#44;&#92;&#44;&#92;&#44;&#92;&#44;&#32;&#102;&#40;&#120;&#41;&#32;&#61;&#32;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#92;&#115;&#105;&#103;&#109;&#97;&#92;&#115;&#113;&#114;&#116;&#123;&#50;&#92;&#112;&#105;&#125;&#125;&#101;&#94;&#123;&#45;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#125;&#40;&#92;&#102;&#114;&#97;&#99;&#123;&#120;&#45;&#92;&#109;&#117;&#125;&#123;&#92;&#115;&#105;&#103;&#109;&#97;&#125;&#41;&#94;&#50;&#125;&#32;&#92;&#101;&#110;&#100;&#32;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#70;&#105;&#110;&#97;&#108;&#108;&#121;&#44;&#32;&#119;&#101;&#32;&#107;&#110;&#111;&#119;&#32;&#116;&#104;&#97;&#116;&#32;&#116;&#104;&#101;&#32;&#109;&#97;&#116;&#104;&#101;&#109;&#97;&#116;&#105;&#99;&#97;&#108;&#32;&#101;&#120;&#112;&#101;&#99;&#116;&#97;&#116;&#105;&#111;&#110;&#32;&#111;&#102;&#32;&#97;&#32;&#114;&#97;&#110;&#100;&#111;&#109;&#32;&#118;&#97;&#114;&#105;&#97;&#98;&#108;&#101;&#32;&#99;&#97;&#110;&#32;&#98;&#101;&#32;&#101;&#115;&#116;&#105;&#109;&#97;&#116;&#101;&#100;&#32;&#112;&#111;&#105;&#110;&#116;&#119;&#105;&#115;&#101;&#32;&#98;&#121;&#58;&#32;&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#32;&#92;&#109;&#117;&#32;&#92;&#115;&#105;&#109;&#101;&#113;&#32;&#92;&#98;&#97;&#114;&#123;&#88;&#125;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#110;&#125;&#92;&#115;&#117;&#109;&#95;&#123;&#105;&#61;&#49;&#125;&#94;&#123;&#110;&#125;&#88;&#95;&#105;&#32;&#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;&#34;&#32;&#116;&#105;&#116;&#108;&#101;&#61;&#34;&#82;&#101;&#110;&#100;&#101;&#114;&#101;&#100;&#32;&#98;&#121;&#32;&#81;&#117;&#105;&#99;&#107;&#76;&#97;&#84;&#101;&#88;&#46;&#99;&#111;&#109;&#34;&#47;&#62; &#83;&#111;&#44;&#32;&#102;&#111;&#114;&#32;&#111;&#117;&#114;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>x<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c598e9d3a603a165dd10ae0011aaaa19_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#101;&#97;&#115;&#117;&#114;&#101;&#44;&#32;&#116;&#104;&#101;&#32;&#115;&#97;&#109;&#112;&#108;&#101;&#115;&#32;&#97;&#114;&#101;&#32;&#97;&#115;&#115;&#117;&#109;&#101;&#100;&#32;&#116;&#111;&#32;&#102;&#111;&#108;&#108;&#111;&#119;&#32;&#97;&#32;&#78;&#111;&#114;&#109;&#97;&#108;&#32;&#100;&#105;&#115;&#116;&#114;&#105;&#98;&#117;&#116;&#105;&#111;&#110;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"496\" style=\"vertical-align: -4px;\"\/>N(\\bar<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dabf1d03e569cb16bc97254adcba7d99_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#41;&#46;&#123;&#120;&#125;&#44;&#92;&#115;&#105;&#103;&#109;&#97;&#94;&#50;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"53\" style=\"vertical-align: -5px;\"\/> whose probability density will be expressed in matrix form:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 45px;\"><span class=\"ql-right-eqno\"> (2.3.1.7) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e843214fc95ac522a916b1d2ac922f78_l3.png\" height=\"45\" width=\"394\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#80;&#40;&#120;&#41;&#61;&#32;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#92;&#112;&#105;&#94;&#123;&#102;&#114;&#97;&#99;&#123;&#110;&#125;&#123;&#50;&#125;&#125;&#124;&#92;&#80;&#104;&#105;&#124;&#94;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#125;&#125;&#92;&#101;&#120;&#112;&#92;&#123;&#45;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#125;&#40;&#120;&#45;&#92;&#98;&#97;&#114;&#123;&#120;&#125;&#41;&#94;&#116;&#91;&#92;&#80;&#104;&#105;&#93;&#94;&#123;&#45;&#49;&#125;&#40;&#120;&#45;&#92;&#98;&#97;&#114;&#123;&#120;&#125;&#41;&#92;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>with <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-20ebc7ea21d727ebe3840c06a433ab09_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#98;&#97;&#114;&#123;&#120;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"10\" style=\"vertical-align: 0px;\"\/> mean of a class, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9cd53f60affa6d7abdebaba9082417ea_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#91;&#92;&#80;&#104;&#105;&#95;&#107;&#93;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"27\" style=\"vertical-align: -5px;\"\/> covariance matrix of class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2865017a3230b92851c401ce3361c750_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#124;&#92;&#80;&#104;&#105;&#95;&#107;&#124;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"27\" style=\"vertical-align: -5px;\"\/> determinant of covariance matrix of class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/>.<\/p>\n<p>The conditional probability is therefore expressed by the following equation:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 45px;\"><span class=\"ql-right-eqno\"> (2.3.1.8) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a24bf3c7e4db7e433beeeba4c1dced1c_l3.png\" height=\"45\" width=\"491\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#80;&#40;&#120;&#124;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;&#61;&#32;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#92;&#112;&#105;&#94;&#123;&#102;&#114;&#97;&#99;&#123;&#110;&#125;&#123;&#50;&#125;&#125;&#124;&#92;&#80;&#104;&#105;&#95;&#107;&#124;&#94;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#125;&#125;&#92;&#101;&#120;&#112;&#92;&#123;&#45;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#50;&#125;&#40;&#120;&#45;&#92;&#98;&#97;&#114;&#123;&#120;&#125;&#95;&#107;&#41;&#94;&#116;&#91;&#92;&#80;&#104;&#105;&#95;&#107;&#93;&#94;&#123;&#45;&#49;&#125;&#40;&#120;&#45;&#92;&#98;&#97;&#114;&#123;&#120;&#125;&#95;&#107;&#41;&#92;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The classification operation consists in calculating for each measure <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/> its probability of belonging to all the classes, and assigning it to the class that verifies the maximum a posteriori probability.<\/p>\n<h4>2.3.2 K-nearest neighbors (KPPV or KNN)<\/h4>\n<p>This approach consists in taking a given number (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/>) of known individuals from the attribute space, individuals chosen from the neighborhood of the individual whose class we&#8217;re trying to estimate. The labeling of the unknown individual is based on knowledge of the classes represented in this neighborhood. <\/p>\n<p>The k-nearest neighbor rule was formalized by Fix and Hodges [5] in the 1950s. This rule consists in basing the choice of the neighborhood <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a6e421454947c585b8fb5ae10299f873_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#98;&#98;&#123;&#82;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"13\" style=\"vertical-align: 0px;\"\/> on the number of observations it contains, fixed at <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> among the total number of observations <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/>. In other words, we select a subset of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> individuals from the set of observations. It is the volume <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3be09045ca01725e04c2484883ac16ac_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#86;&#95;&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"17\" style=\"vertical-align: -3px;\"\/> of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a6e421454947c585b8fb5ae10299f873_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#98;&#98;&#123;&#82;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"13\" style=\"vertical-align: 0px;\"\/> that is variable. The probability density estimator is then expressed as follows:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 95px;\"><span class=\"ql-right-eqno\"> (2.3.2.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-06127c407f397954640efd1081c170eb_l3.png\" height=\"95\" width=\"728\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#104;&#97;&#116;&#123;&#80;&#125;&#40;&#120;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#107;&#125;&#123;&#109;&#92;&#116;&#105;&#109;&#101;&#115;&#32;&#86;&#95;&#107;&#125; &#92;&#101;&#110;&#100;&#32;&#32;&#32;&#32;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#84;&#104;&#105;&#115;&#32;&#101;&#115;&#116;&#105;&#109;&#97;&#116;&#101;&#32;&#105;&#115;&#32;&#118;&#97;&#108;&#105;&#100;&#32;&#111;&#110;&#108;&#121;&#32;&#117;&#110;&#100;&#101;&#114;&#32;&#116;&#104;&#101;&#32;&#102;&#111;&#108;&#108;&#111;&#119;&#105;&#110;&#103;&#32;&#99;&#111;&#110;&#100;&#105;&#116;&#105;&#111;&#110;&#115;&#58; &#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#42;&#125; &#92;&#108;&#101;&#102;&#116;&#92;&#123; &#92;&#98;&#101;&#103;&#105;&#110;&#123;&#97;&#114;&#114;&#97;&#121;&#125;&#123;&#99;&#99;&#99;&#125; &#92;&#108;&#105;&#109;&#95;&#123;&#109;&#92;&#114;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#32;&#43;&#92;&#105;&#110;&#102;&#116;&#121;&#125;&#86;&#95;&#107;&#38;&#61;&#38;&#48;&#92; &#92;&#108;&#105;&#109;&#95;&#123;&#109;&#92;&#114;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#32;&#43;&#92;&#105;&#110;&#102;&#116;&#121;&#125;&#92;&#102;&#114;&#97;&#99;&#123;&#107;&#125;&#123;&#86;&#95;&#107;&#125;&#38;&#61;&#38;&#48; &#92;&#101;&#110;&#100;&#123;&#97;&#114;&#114;&#97;&#121;&#125; &#92;&#114;&#105;&#103;&#104;&#116;&#46; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The first assumption corresponds to a space completely occupied by observations: when the number of observations is infinite, these observations occupy the space uniformly. The second hypothesis considers that, for an infinite number of observations, any subset of cardinal <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> is occupied by an infinite number of uniformly distributed elements. <\/p>\n<p>In the context of supervised classification, the conditional probability distribution can be approximated by:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 42px;\"><span class=\"ql-right-eqno\"> (2.3.2.2) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b47c0dd6f1fed704989c2227de190755_l3.png\" height=\"42\" width=\"144\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#104;&#97;&#116;&#123;&#80;&#125;&#40;&#120;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#107;&#95;&#113;&#125;&#123;&#109;&#95;&#113;&#92;&#116;&#105;&#109;&#101;&#115;&#32;&#65;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ac50939091356f1b1e4313c3b58bd26a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"16\" style=\"vertical-align: -6px;\"\/> is the number of observations contained in the volume <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3be09045ca01725e04c2484883ac16ac_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#86;&#95;&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"17\" style=\"vertical-align: -3px;\"\/> belonging to the label class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d9121146a128b872d348446882fb8bdd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"23\" style=\"vertical-align: -6px;\"\/> the total number of observations belonging to the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/>. The a posteriori law of observation of a label conditional on an observation is obtained with Bayes&#8217; rule: <\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 55px;\"><span class=\"ql-right-eqno\"> (2.3.2.3) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-52d57d249f2fd9b6d49bcb13db6d9ae0_l3.png\" height=\"55\" width=\"502\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#104;&#97;&#116;&#123;&#80;&#125;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#121;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;&#92;&#104;&#97;&#116;&#123;&#80;&#125;&#40;&#121;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;&#125;&#123;&#92;&#115;&#117;&#109;&#95;&#107;&#32;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;&#92;&#104;&#97;&#116;&#123;&#80;&#125;&#40;&#121;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;&#125;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;&#92;&#102;&#114;&#97;&#99;&#123;&#107;&#95;&#113;&#125;&#123;&#109;&#95;&#113;&#116;&#105;&#109;&#101;&#115;&#32;&#86;&#95;&#107;&#125;&#125;&#123;&#104;&#97;&#116;&#123;&#80;&#125;&#40;&#121;&#41;&#125;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;&#92;&#116;&#105;&#109;&#101;&#115;&#32;&#109;&#125;&#123;&#109;&#95;&#113;&#125;&#92;&#102;&#114;&#97;&#99;&#123;&#107;&#95;&#113;&#125;&#123;&#107;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The choice of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> is linked to the learning base. Choosing a large <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> allows local properties to be exploited to the full, but requires a large number of samples to keep the neighborhood volume small. Often, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> is chosen as the square root of the average number of elements per class, i.e. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1bcf3d02d9135ea80af43a2112f490f8_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#115;&#113;&#114;&#116;&#123;&#92;&#102;&#114;&#97;&#99;&#123;&#109;&#125;&#123;&#99;&#125;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"22\" width=\"34\" style=\"vertical-align: -7px;\"\/> [2].  <\/p>\n<h3>2.4 Summary<\/h3>\n<p>But how do you choose anything other than what is statistically most likely?<\/p>\n<p>The problem then arises of measuring the accuracy of a decision and producing a measure of accuracy, even if the problem may be this measure.<\/p>\n<h2>3 Theory of evidence<\/h2>\n<p>The transferable belief model (TBM) is a generic formal framework developed by Philippe Smets [20] for the representation and combination of knowledge. TBM is based on the definition of belief functions provided by information sources that may be complementary, redundant and possibly non-independent. It proposes a set of operators for combining these functions. It is therefore naturally used in the context of information fusion to improve the analysis and interpretation of data from multiple information sources. This framework naturally corresponds to the one we&#8217;re concerned with in its decision-making aspect, thanks to different information from different attributes.    <\/p>\n<p>One of the fundamental features of TBM is the differentiation between the knowledge representation and decision levels. This differentiation is much less preponderant in other approaches, particularly the probabilistic model, where decision is often the sole objective. TBM&#8217;s reasoning mechanisms are therefore grouped into two levels, as illustrated in figure 3.1 :  <\/p>\n<ul>\n<li>The <strong>credal<\/strong> level: home to knowledge representation (static part), combinations and reasoning about this knowledge (dynamic part).<\/li>\n<li>The <strong>pignistic<\/strong> level indicates a decision, possibly taking into account the risk and\/or reward associated with that decision.<\/li>\n<\/ul>\n<figure id=\"attachment_3249\" style=\"width: 600px; margin-left: auto; margin-right: auto;\"><img loading=\"lazy\" decoding=\"async\" class=\"size-medium wp-image-3249\" src=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT-300x116.png\" alt=\"MCT\" width=\"600\" height=\"232\" srcset=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT-300x116.png 300w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT-1024x395.png 1024w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT-768x296.png 768w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT-1536x592.png 1536w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT-1568x604.png 1568w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/MCT.png 1892w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/><figcaption>Figure 3.1 &#8211; Mechanisms of the transferable belief model<\/figcaption><\/figure>\n<p>In the context of representation in the form of a processing chain, we are faced with the problem of observing the same event from several sources, and combining the information provided by these sources (see Figure 3.2), which raises the following questions:<\/p>\n<ul>\n<li>How can we represent the knowledge of an information source in the form of belief functions, and what advantages can we derive from such a representation?<\/li>\n<li>How can we combine several sources of belief functions to summarize information and improve decision-making?<\/li>\n<li>How to make a decision based on belief functions?<\/li>\n<\/ul>\n<figure id=\"attachment_3251\" style=\"width: 600px; margin-left: auto; margin-right: auto;\"><img loading=\"lazy\" decoding=\"async\" class=\"size-medium wp-image-3251\" src=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion-300x126.png\" alt=\"merger\" width=\"600\" height=\"252\" srcset=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion-300x126.png 300w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion-1024x429.png 1024w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion-768x322.png 768w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion-1536x643.png 1536w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion-1568x657.png 1568w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/fusion.png 1979w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/><figcaption>Figure 3.2 &#8211; Processing chain<\/figcaption><\/figure>\n<h3>3.1 Presentation<\/h3>\n<p>Let&#8217;s take the example of handwritten number recognition. The possible digits that can be written are called <strong>hypotheses<\/strong>, and the set of hypotheses forms the <strong>discernment frame<\/strong>, usually denoted <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>. For our example, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a21ee6662cf56bb9b887369dd5e9dd89_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;&#61;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#51;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#52;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#53;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#54;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#56;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#57;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"311\" style=\"vertical-align: -5px;\"\/> where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d2a183e185073bbe1bd87f051d82b9b3_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"16\" style=\"vertical-align: -3px;\"\/> is the symbol representing the digit <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-695d9d59bd04859c6c99e7feb11daab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"6\" style=\"vertical-align: 0px;\"\/>. Since two hypotheses can&#8217;t be true simultaneously (a symbol representing several digits can&#8217;t be written at the same time), they are assumed to be <g id=\"gid_2\">exclusive<\/g>, resulting in a null intersection between them (i.e. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-390a752c260b75733ab90d25eef70527_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#92;&#125;&#92;&#99;&#97;&#112;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#51;&#92;&#125;&#61;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"163\" style=\"vertical-align: -5px;\"\/>). Now let&#8217;s suppose that the document we&#8217;re studying is degraded, or that the figure is very badly written, a reader would then give a weighted opinion on this figure, making it a <strong>source of<\/strong> information. With his available knowledge, the reader would say that it&#8217;s either <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-55acbef395b67f334d09e534db212b1d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"17\" style=\"vertical-align: -3px;\"\/>, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-4f3c032b406bd361e15ce341b4270129_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> or <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0f7a5d1eea7209f6f6a98bbf19d31779_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#57;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/>. The question is how to represent this information.      <\/p>\n<p>In the context of probability theory, the answer is given by the principle <strong>of equiprobability<\/strong>, i.e. each digit is assigned the same probability (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-34f0e870957984f6c69249b8cf4f5813_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#51;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"22\" width=\"7\" style=\"vertical-align: -6px;\"\/>) so that the sum of all probabilities is equal to 1. Thus, the probability distribution for this digit is equal to:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 36px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f2b3ea0be73fc4930a94e59a35d5907f_l3.png\" height=\"36\" width=\"221\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#41;&#61;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#41;&#61;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#57;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#51;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>This probability is only assigned to hypotheses taken alone (also called <i>singletons<\/i>). <\/p>\n<p>In evidence theory, the main notion is that of <strong>belief functions<\/strong>, within which knowledge is modeled by a distribution of <strong>belief masses<\/strong>. In our example, the entire belief mass (an integer unit, as with probabilities) is assigned to the union of all hypotheses, i.e. the <strong>proposition<\/strong> <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e9fd12c53b2bfa99556373223c074dcf_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#57;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"128\" style=\"vertical-align: -5px;\"\/>:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 19px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-76c73d5d93da6a34b63849d032af7d8f_l3.png\" height=\"19\" width=\"191\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#109;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#57;&#92;&#125;&#41;&#61;&#49; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>As we can see, this mass gives us no information about the belief of each of the hypotheses (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-55acbef395b67f334d09e534db212b1d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"17\" style=\"vertical-align: -3px;\"\/>, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-4f3c032b406bd361e15ce341b4270129_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0f7a5d1eea7209f6f6a98bbf19d31779_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#57;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/>) that make it up. We can then say that the belief mass explicitly models doubt (or ignorance) about this set of hypotheses. To return to probability theory, the value of the probability on a union of hypotheses is implicitly derived from the probability on the singletons, a probability established from the sum of all the private probabilities of the common part, i.e. the probability of the intersection. For example:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 19px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dfb009a37adf1e116db9a110a047b635_l3.png\" height=\"19\" width=\"395\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#80;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#92;&#125;&#41;&#61;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#41;&#43;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#41;&#45;&#80;&#40;&#92;&#123;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#41;&#92;&#125;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>Here, the probability on <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-67eb8da78308a9f115ec1887607a60fe_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#92;&#123;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#41;&#92;&#125;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"104\" style=\"vertical-align: -5px;\"\/> represents the common part of the events<i>&#8220;The number is 1<\/i>&#8221; and<i>&#8220;The number is 7<\/i>&#8220;, which we need to remove so as not to count the events twice. In the case where the hypotheses are exclusive (a case we&#8217;ll come back to throughout this article), we obtain :<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 19px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dbf615cb1483b658cc36c5d271e5098f_l3.png\" height=\"19\" width=\"631\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#125;&#92;&#99;&#97;&#112;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#125;&#61;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;&#32;&#92;&#82;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#32;&#80;&#40;&#92;&#123;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#41;&#92;&#125;&#41;&#61;&#48;&#32;&#92;&#82;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#32;&#80;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#92;&#125;&#41;&#61;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#41;&#43;&#80;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#55;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>taking into account that <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-796ccdda5ba50e5b6ac99a29b379c172_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#80;&#40;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;&#41;&#61;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"69\" style=\"vertical-align: -5px;\"\/> by definition. <\/p>\n<p>Unlike probabilities, a mass function on a union of hypotheses is not equal to the sum of the masses of the hypotheses making up the union [10], and it is precisely thanks to this property that we represent doubt between hypotheses. Moreover, belief functions have the advantage of being generic, since they are the mathematical representation of a superset of probability functions. Indeed, a mass distribution, where only singleton hypotheses have non-zero mass, can be interpreted as a probability distribution. Klir and Wierman explain in [10] that belief functions mathematically generalize possibility functions, and that each of these functions (probabilities, possibilities and beliefs) has particular characteristics and enables knowledge to be modeled differently.   <\/p>\n<p>Let&#8217;s now look at the mathematical formalism of belief functions, based on the models of Dempster [6][7], Shafer [15], transferable beliefs [20] and the Hints model [9]. For this work, we have restricted ourselves to the axiomatized and formalized framework of the transferable belief model developed by P. Smets. <\/p>\n<h3>3.2 Formalism<\/h3>\n<h4>3.2.1 Introduction<\/h4>\n<p>If we return to the problem of managing multiple sources of information (as illustrated in Figure 3.2), what we&#8217;re trying to do is determine the actual <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-88879811a973dafad289e3bdc75abf6e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> state of the system under study, using multiple observations from multiple observers. In our case, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-88879811a973dafad289e3bdc75abf6e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> will be the decision to be produced and the various attributes provide the observations on the system. <\/p>\n<p>We assume that this state <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-88879811a973dafad289e3bdc75abf6e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> takes discrete values <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e5df6ddf8612587097c509ab7073284f_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> (the <strong>hypotheses<\/strong>) and that the set of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5793832f979c2268e3694c246d53b1bb_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#78;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"16\" style=\"vertical-align: 0px;\"\/> possible hypotheses is called the <strong>discernment frame<\/strong> (or \\textbf{universe of discourse}), generally noted:<br \/>\n<a name=\"id966678636\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 53px;\"><span class=\"ql-right-eqno\"> (3.2.1.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b6ebf3741b40ec2f0048626239480cb_l3.png\" height=\"53\" width=\"263\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#79;&#109;&#101;&#103;&#97;&#61;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#46;&#46;&#46;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#44;&#46;&#46;&#46;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#78;&#92;&#125;&#61;&#92;&#98;&#105;&#103;&#99;&#117;&#112;&#95;&#123;&#107;&#61;&#49;&#125;&#94;&#123;&#78;&#125;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The total number of hypotheses making up <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> is called the cardinal and is written as <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b8061f11cb1c8c3cb6609178d22494df_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#124;&#92;&#79;&#109;&#101;&#103;&#97;&#124;&#61;&#78;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"60\" style=\"vertical-align: -5px;\"\/>. As we saw earlier, all hypotheses are considered exclusive. <\/p>\n<p>The system&#8217;s set of observers (the sensors in figure 3.2) provides a weighted opinion, represented by a belief function, about its actual state <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d37df2ddca754210f83ee823df58c25d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#111;&#109;&#101;&#103;&#97;&#95;&#48;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#32;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"94\" style=\"vertical-align: -4px;\"\/>. We call these sensors belief <strong>sources<\/strong>. <\/p>\n<h4>3.2.2 Mass functions<\/h4>\n<p>Let <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2b56ee3ef0153878226308be22d08790_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#50;&#94;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"19\" style=\"vertical-align: 0px;\"\/> be the space formed by all the parts of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>, i.e. the space containing all the possible subsets formed by the hypotheses and unions of hypotheses, i.e. :<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 23px;\"><span class=\"ql-right-eqno\"> &nbsp; <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6d78ce96afc43d916a3cfe9ada8a4fd0_l3.png\" height=\"23\" width=\"636\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#50;&#94;&#92;&#79;&#109;&#101;&#103;&#97;&#61;&#92;&#123;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#92;&#125;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#50;&#92;&#125;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;&#92;&#125;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#51;&#92;&#125;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#51;&#92;&#125;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#51;&#92;&#125;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#51;&#92;&#125;&#44;&#46;&#46;&#46;&#44;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#46;&#46;&#46;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#78;&#92;&#125;&#92;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>Let <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> be a <strong>proposition<\/strong> such that <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-28e1f1a318e2d41bc60fab4658efb076_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"49\" style=\"vertical-align: -3px;\"\/>, e.g. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f6e9c648eb9aabc5fb470c48ed311ece_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#61;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"99\" style=\"vertical-align: -5px;\"\/>. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> explicitly represents the doubt between its hypotheses (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-55acbef395b67f334d09e534db212b1d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"17\" style=\"vertical-align: -3px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8c5d046cd42cadba042af2cbde0cca9c_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#50;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/>). As we saw above, belief masses are non-additive, which is a fundamental difference from probability theory. Thus, the belief mass <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fd406af8e3a5895bdf679b2c14ae95ae_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"42\" style=\"vertical-align: -5px;\"\/> allocated to <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> gives no information about the hypotheses and subsets making up <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-25b206f25506e6d6f46be832f7119ffa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: 0px;\"\/> [10].   <\/p>\n<p>We can define a <strong>distribution of belief masses<\/strong> (BBA: basic belief assignment) as a set of belief masses concerning any <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-28e1f1a318e2d41bc60fab4658efb076_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"49\" style=\"vertical-align: -3px;\"\/> propositions verifying:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 40px;\"><span class=\"ql-right-eqno\"> (3.2.2.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-92a7b1a13b47b2b97b617ec61b1141bf_l3.png\" height=\"40\" width=\"109\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#115;&#117;&#109;&#95;&#123;&#65;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#79;&#109;&#101;&#103;&#97;&#125;&#109;&#40;&#65;&#41;&#61;&#49; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>Each of the sensors observing the system will contribute its own mass distribution.<\/p>\n<p>The mass <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fd406af8e3a5895bdf679b2c14ae95ae_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#65;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"42\" style=\"vertical-align: -5px;\"\/> supplied by a sensor is that sensor&#8217;s share of belief in the proposition &#8220;<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0eb67e4373d5dc6763292153a1d77ebb_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;&#92;&#105;&#110;&#32;&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"54\" style=\"vertical-align: -3px;\"\/>&#8220;, where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-88879811a973dafad289e3bdc75abf6e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> is the actual state of the observed system. Any non-zero mass element <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> (i.e. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6f8b68d800ae75748e8615a5910674e5_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#66;&#41;&#62;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"76\" style=\"vertical-align: -5px;\"\/>) is called the <strong>focal element<\/strong> of the belief mass distribution. Ramasso [13] presents a notable set of belief mass distributions in his dissertation. <\/p>\n<p>When the universe of discourse contains all possible hypotheses (i.e. when it is <strong>exhaustive<\/strong>), it necessarily contains the hypothesis that describes the observed system. Otherwise, the discernment framework is not adapted to the problem at hand. <\/p>\n<p>From the moment <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> is exhaustive, the mass of the empty set is zero (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c33fab0285ffd99010ed9e367e3cd50e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;&#41;&#61;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"71\" style=\"vertical-align: -5px;\"\/>), the mass is said to be <strong>normal<\/strong>, and the problem then enters the framework of a <strong>closed world<\/strong> as in the case of Shafer&#8217;s model [15].<\/p>\n<p>In the case of the transferable belief model, the universe of discourse can be non-exhaustive; the mass of the set is non-zero (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-569a702319260518e458b455009c981f_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#123;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;&#125;&#41;&#92;&#110;&#101;&#113;&#32;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"71\" style=\"vertical-align: -5px;\"\/>) and is said to be <strong>subnormal<\/strong>, and in this case the problematic is part of an <strong>open-world<\/strong> framework. This type of construction makes it possible to redefine the discernment framework, for example, by adding a hypothesis based on the mass of the empty set, such as a <strong>rejection hypothesis<\/strong>. Rejection hypotheses correspond to states such as<i>&#8220;I don&#8217;t know<\/i>&#8220;. At the end of the chain, they allow us to reject an individual for whom there is too much doubt, rather than taking the risk of classifying him or her in spite of everything. To move from an open to a closed world, we need to redistribute the mass of the empty set to the other subsets of the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2b56ee3ef0153878226308be22d08790_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#50;&#94;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"19\" style=\"vertical-align: 0px;\"\/> space [11].   <\/p>\n<p>To summarize what we&#8217;ve seen so far, the mass function is determined a priori, without the need for probabilistic considerations. It may correspond to an observer&#8217;s subjective attribution of degrees of credibility to the realization of various conceivable events on <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>. <\/p>\n<p>The assignment of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/> values is used to express the degree to which the expert or observer judges that each hypothesis is likely to come true. Thus, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-607bbc78fe6b69f04d30f8518937f835_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"42\" style=\"vertical-align: -5px;\"\/> describes its imprecision relative to these degrees. <\/p>\n<h4>3.2.3 Dempster&#8217;s combination rule<\/h4>\n<p>If we return to figure 3.2, we see that the problem at hand is information fusion, i.e. <em>combining heterogeneous information from several sources to improve decision-making<\/em> [3].<\/p>\n<p>Let&#8217;s now look at a formal tool developed within the framework of the transferable belief model for information fusion. We assume that the sources are <strong>distinct<\/strong> [16][21][22][23], by analogy with the<strong>independence<\/strong> of random variables in statistics. <\/p>\n<p>It may happen that for the same hypothesis, two sensors provide completely different information. From a formal point of view, when combining two belief mass functions <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-51921237944fd6e43f0640228a37376f_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"22\" style=\"vertical-align: -3px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f6dae86895dac0d4644151786b47c7ce_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#50;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"23\" style=\"vertical-align: -3px;\"\/>, the intersection between the focal elements may be empty. The resulting belief mass function in this combination will be non-zero on the empty set. This value then quantifies the discordance between the belief sources, and is called <strong>conflict<\/strong> [17]. A normalization can be performed, bringing the conjunctive combination rule back to <strong>Dempster&#8217;s orthogonal rule<\/strong> [6][7] (denoted <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2dd05ae7ff8e9f20d4decb554faeac61_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#112;&#108;&#117;&#115;\" title=\"Rendered by QuickLaTeX.com\" height=\"13\" width=\"13\" style=\"vertical-align: -2px;\"\/>) used in Shafer&#8217;s model [15][18]:<br \/>\n\\begin    {eqnarray}<br \/>\nm_{1\\oplus 2}(A)=(m_1 \\oplus m_2)(A)=\\left\\{<br \/>\n\\begin{array}{lccc}<br \/>\n\\frac{m_{1\\overlay{\\cap}{\\bigcirc}2}(A)}{1-m_{1\\overlay{\\cap}{\\bigcirc}2}(\\emptyset)}, &amp; \\forall A\\subseteq\\Omega &amp; \\mathrm{si }&amp;A\\neq\\emptyset\\\\\\<br \/>\n0 &amp; &amp; \\mathrm{si }&amp;A=\\emptyset\\\\<br \/>\n\\end{array}<br \/>\n\\right.<br \/>\n\\label{incertitude:eq:dempsterrule}\\tag{3.2.3.1}<br \/>\n\\end{eqnarray}<br \/>\nAnother way of writing this combination rule is:<\/p>\n<pre class=\"ql-errors\">*** QuickLaTeX cannot compile formula:\n&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\n&#109;&#40;&#65;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#123;&#115;&#117;&#109;&#95;&#123;&#66;&#92;&#99;&#97;&#112;&#32;&#67;&#61;&#65;&#125;&#109;&#95;&#49;&#40;&#66;&#41;&#109;&#95;&#50;&#40;&#67;&#41;&#125;&#123;&#123;&#115;&#117;&#109;&#95;&#123;&#66;&#92;&#99;&#97;&#112;&#32;&#67;&#110;&#101;&#113;&#48;&#125;&#109;&#95;&#49;&#40;&#66;&#41;&#109;&#95;&#50;&#40;&#67;&#41;&#125;&#32;&#38;&#32;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#65;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#79;&#109;&#101;&#103;&#97;\n&#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\n\n*** Error message:\n&#70;&#105;&#108;&#101;&#32;&#101;&#110;&#100;&#101;&#100;&#32;&#119;&#104;&#105;&#108;&#101;&#32;&#115;&#99;&#97;&#110;&#110;&#105;&#110;&#103;&#32;&#117;&#115;&#101;&#32;&#111;&#102;&#32;&#92;&#102;&#114;&#97;&#99;&#32;&#46;\r\n&#69;&#109;&#101;&#114;&#103;&#101;&#110;&#99;&#121;&#32;&#115;&#116;&#111;&#112;&#46;\r\n\n<\/pre>\n<p>With, as a measure of the conflict between sources:<br \/>\n<a name=\"id1901337897\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 38px;\"><span class=\"ql-right-eqno\"> (3.2.3.3) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ea97d9b6f4f052124b20f819774247bb_l3.png\" height=\"38\" width=\"193\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#75;&#32;&#61;&#32;&#92;&#115;&#117;&#109;&#95;&#123;&#66;&#92;&#99;&#97;&#112;&#32;&#67;&#61;&#48;&#125;&#109;&#95;&#49;&#40;&#66;&#41;&#109;&#95;&#50;&#40;&#67;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<h4>3.2.4 Simple attenuation<\/h4>\n<p>The <strong>simple weakening rule<\/strong> to incorporate source reliability. Applying the disjunctive combination rule can be too conservative when the sources of belief mass functions are unreliable. It can thus lead to a completely non-informative (empty) belief mass function. In some applications, however, it is possible to quantify reliability, enabling a less conservative combination rule to be applied in the case of unreliable sources.   <\/p>\n<p>Taking reliability into account in the context of belief functions is called<strong>weakening<\/strong>, because the process consists in weighting the mass of elements of a belief mass function. The first work on weakening in the context of belief functions was developed by Shafer [15], axiomatized by Smets [16] and generalized by Mercier and Den\u0153ux [12]. Generalized weakening was called contextual weakening.  <\/p>\n<p>Shafer&#8217;s simple weakening of a belief mass function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/> is defined as the weighting of each mass <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-73899e3fe6635e5630b38a80684c6438_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#40;&#66;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"43\" style=\"vertical-align: -5px;\"\/> of the distribution by a coefficient <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-246f9207961a8e475122d388322c9222_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#40;&#49;&#45;&#92;&#97;&#108;&#112;&#104;&#97;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"54\" style=\"vertical-align: -5px;\"\/> called reliability and where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8d1a0bdc10aa5d9a9cd33815a79df449_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#97;&#108;&#112;&#104;&#97;&#92;&#105;&#110;&#91;&#48;&#44;&#49;&#93;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"67\" style=\"vertical-align: -5px;\"\/> is the rate of weakening.<\/p>\n<p>Formally:<br \/>\n<a name=\"id987005651\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 45px;\"><span class=\"ql-right-eqno\"> (3.2.4.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1858061df79dfe737af46efc09167eca_l3.png\" height=\"45\" width=\"301\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#109;&#94;&#92;&#97;&#108;&#112;&#104;&#97;&#40;&#66;&#41;&#32;&#61;&#38;&#32;&#40;&#49;&#45;&#92;&#97;&#108;&#112;&#104;&#97;&#41;&#46;&#109;&#40;&#66;&#41;&#44;&#38;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#66;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#92;&#79;&#109;&#101;&#103;&#97;&#92;&#92; &#109;&#94;&#92;&#97;&#108;&#112;&#104;&#97;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#32;&#61;&#38;&#32;&#40;&#49;&#45;&#92;&#97;&#108;&#112;&#104;&#97;&#41;&#46;&#109;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#43;&#92;&#97;&#108;&#112;&#104;&#97; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>By the principle of minimum information (maximum entropy), the rest of the mass (after weighting) is transferred to the element of ignorance <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>. The more reliable the source (and the lower the rate of weakening <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8f0b6b1a01f8fcc2f95be0364c090397_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#97;&#108;&#112;&#104;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"11\" style=\"vertical-align: 0px;\"\/>), the less the belief mass function is modified, whereas as reliability decreases, the belief mass function resulting from weakening tends towards the empty belief mass function. <\/p>\n<h3>3.3 The pignistic decision<\/h3>\n<p>From the set of belief functions provided by all the sensors, and following their combination by Dempster&#8217;s rule, we obtain a unique belief function for each of the possible hypotheses modeling the system&#8217;s knowledge. This knowledge is then used to make a decision on one of the hypotheses. <\/p>\n<p>Note that the system&#8217;s hypotheses don&#8217;t necessarily represent the only possible classes (i.e. the 26 letters of our alphabet). It is also possible to create a new hypothesis corresponding to the event &#8220;none of the known hypotheses&#8221;. This is referred to as <g id=\"gid_0\">rejection<\/g>, i.e. the unknown individual is not re-injected into the chain, but simply classified as &#8220;impossible to recognize&#8221;. The measure of sensor conflict during the Dempster merge (see equation 3.2.3.3) is also a measure on which the system can base its rejection of the individual.   <\/p>\n<p>Information fusion systems, whether based on probability theory, possibility theory or belief functions, are designed for decision making and decision analysis. Decision-making consists in choosing a hypothesis on a <strong>betting framework<\/strong>, generally the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> discernment framework. Decision-making can be automatic or left to the responsibility of the end-user (as in the case of diagnostic aids in the field of medicine).  <\/p>\n<p>Within the framework of the transferable belief model, the decision phase is based on the pignistic probability distribution noted <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-65ee269574210bf52438e31dce0756ad_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#66;&#101;&#116;&#80;&#125;&#92;&#123;&#109;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"72\" style=\"vertical-align: -5px;\"\/> from the mass distribution <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/> [19]. The pignistic transform consists of distributing the mass of a proposition <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> equiprobably over the hypotheses contained in <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-770fd1447ccf2fc229801b486b0d8f8a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#66;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>. <\/p>\n<p>Formally:<br \/>\n<a name=\"id3949396658\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 45px;\"><span class=\"ql-right-eqno\"> (3.3.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-93f1794c42fb2d3438331f46be34ca17_l3.png\" height=\"45\" width=\"239\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#66;&#101;&#116;&#125;&#92;&#123;&#109;&#92;&#125;&#58;&#32;&#38;&#32;&#92;&#79;&#109;&#101;&#103;&#97;&#92;&#114;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#91;&#48;&#44;&#49;&#93;&#92;&#92; &#38;&#32;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#109;&#97;&#112;&#115;&#116;&#111;&#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#66;&#101;&#116;&#80;&#125;&#92;&#123;&#109;&#92;&#125;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d31f9c95481a703630a82d89e73fb894_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#66;&#101;&#116;&#80;&#125;&#92;&#123;&#109;&#92;&#125;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;&#44;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#105;&#110;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"177\" style=\"vertical-align: -5px;\"\/> is given by:<br \/>\n<a name=\"id2451275286\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 50px;\"><span class=\"ql-right-eqno\"> (3.3.2) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d468522a0ceadfec8f71bde6e45152d5_l3.png\" height=\"50\" width=\"374\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#66;&#101;&#116;&#80;&#125;&#92;&#123;&#109;&#92;&#125;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;&#61;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#40;&#49;&#45;&#109;&#40;&#92;&#101;&#109;&#112;&#116;&#121;&#115;&#101;&#116;&#41;&#41;&#125;&#92;&#115;&#117;&#109;&#95;&#123;&#66;&#92;&#115;&#117;&#98;&#115;&#101;&#116;&#101;&#113;&#92;&#92;&#79;&#109;&#101;&#103;&#97;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#105;&#110;&#32;&#66;&#125;&#92;&#102;&#114;&#97;&#99;&#123;&#109;&#40;&#66;&#41;&#125;&#123;&#124;&#66;&#124;&#125; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The decision is generally made by choosing the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e5df6ddf8612587097c509ab7073284f_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> element with the highest pignistic probability:<br \/>\n<a name=\"id353096718\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 28px;\"><span class=\"ql-right-eqno\"> (3.3.3) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9ed849548cadfa07ede0154849e3b893_l3.png\" height=\"28\" width=\"213\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#61;&#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#97;&#114;&#103;&#125;&#92;&#109;&#97;&#120;&#95;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#105;&#110;&#92;&#79;&#109;&#101;&#103;&#97;&#125;&#92;&#116;&#101;&#120;&#116;&#114;&#109;&#123;&#66;&#101;&#116;&#80;&#125;&#92;&#123;&#109;&#92;&#125;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The agent who relies on the pignistic transform in the decision-making phase exhibits rational behavior by maximizing expected utility. Alternatively, plausibilities and credibilities can be used when the agent presents a rather optimistic or pessimistic attitude. <\/p>\n<p>In the case of plausibilities, the agent is said to have an optimistic attitude in the sense that the plausibility of an element represents the maximum value of the belief mass on this element that could be reached if additional information were to reach the fusion system (the latter using the conjunctive combination to integrate this new information). The probability distribution over singletons from which a decision can be made is then given <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-217b01318a8231acbec5781083856bd6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#105;&#110;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"63\" style=\"vertical-align: -3px;\"\/> by:<br \/>\n\\begin {equation}<br \/>\n\\textrm{PIP}\\{m\\}(\\omega_k)=\\frac{pl(\\omega_k)}{\\sum_{\\omega_k\\in\\Omega}pl(\\omega_k)}<br \/>\n\\label{incertitude:eq:pip}\\tag{3.3.4}<br \/>\n\\end{equation}<\/p>\n<p>Note that the plausibility of a singleton <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0796250011db866390ddae2dd7f3d198_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#108;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"48\" style=\"vertical-align: -5px;\"\/> is equal to its communality <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-35e6cfc22467127bd70c616d0d5b63b8_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#113;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"41\" style=\"vertical-align: -5px;\"\/>. In the same way, a criterion adapted to a &#8220;pessimistic&#8221; agent can be obtained using credibilities. The credibility of a singleton <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-7415c3840198544407fd2ae3ad5d4186_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#98;&#101;&#108;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"54\" style=\"vertical-align: -5px;\"\/> is simply equal to its mass after normalization of the belief mass function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/>. Note that for any element <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-7a6a8ac79089691e2f334b6339a7b219_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#92;&#105;&#110;&#50;&#94;&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"17\" width=\"54\" style=\"vertical-align: -1px;\"\/>:<br \/>\n\\begin   {equation}<br \/>\nbel(A)\\leq \\textrm{BetP(A)}\\leq pl(A)<br \/>\n\\label{incertitude:eq:bel}\\tag{3.3.5}<br \/>\n\\end{equation}<br \/>\nand in particular when plausibilities and credibilities are calculated on singletons.<\/p>\n<p>For the decision-making problem, we assume we have a belief function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/> on <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> which summarizes all the information provided on the value of the variable <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0af556714940c351c933bba8cf840796_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#121;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: -4px;\"\/>. The decision consists in choosing an action <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5c53d6ebabdbcfa4e107550ea60b1b19_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"9\" style=\"vertical-align: 0px;\"\/> from a finite set of actions <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2da3234c3fdae205092efb71f0fb7572_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"15\" style=\"vertical-align: -1px;\"\/>. A loss function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-fceb4976674d80b766e6d5ace2f1c199_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#108;&#97;&#109;&#98;&#100;&#97;&#58;&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;&#92;&#116;&#105;&#109;&#101;&#115;&#92;&#79;&#109;&#101;&#103;&#97;&#92;&#114;&#105;&#103;&#104;&#116;&#97;&#114;&#114;&#111;&#119;&#92;&#109;&#97;&#116;&#104;&#98;&#98;&#123;&#82;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"114\" style=\"vertical-align: -1px;\"\/> is assumed to be defined such that <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-af112061d9e21e2363148fd51b81a9d2_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#108;&#97;&#109;&#98;&#100;&#97;&#40;&#97;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"52\" style=\"vertical-align: -5px;\"\/> represents the loss incurred if action <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5c53d6ebabdbcfa4e107550ea60b1b19_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"9\" style=\"vertical-align: 0px;\"\/> is chosen when <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0ff4c6e5fed9da9129267283051dbd64_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#121;&#61;&#92;&#111;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"44\" style=\"vertical-align: -4px;\"\/>. From the pignistic probability, each action <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b6bf7fad7262250a313d692e9a85f969_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;&#92;&#105;&#110;&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"46\" style=\"vertical-align: -1px;\"\/> can be associated with a risk defined as the expected cost relative to <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6db025dc5665e40d45660ab79f5cc73d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#95;&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"22\" style=\"vertical-align: -4px;\"\/> if <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5c53d6ebabdbcfa4e107550ea60b1b19_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"9\" style=\"vertical-align: 0px;\"\/> is chosen:<br \/>\n\\begin   {equation}<br \/>\nR(a)=sum_{\\omega\\in\\Omega}\\lambda(a,\\omega)p_m(\\omega)<br \/>\n\\label{incertitude:eq:risk}\\tag{3.3.6}<br \/>\n\\end{equation}<\/p>\n<p>The idea is then to choose the action that minimizes this risk, generally referred to as pignistic risk. In pattern recognition, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-7f2677ea44715220dca49002df06dee0_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;&#61;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#49;&#44;&#46;&#46;&#46;&#44;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#81;&#92;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"125\" style=\"vertical-align: -6px;\"\/> is the set of classes and the elements of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2da3234c3fdae205092efb71f0fb7572_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#109;&#97;&#116;&#104;&#99;&#97;&#108;&#123;&#65;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"15\" style=\"vertical-align: -1px;\"\/> are generally the actions <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1467057d088a2e895b02c2afeecd8c5e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"16\" style=\"vertical-align: -6px;\"\/> which consist in assigning the unknown vector to the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/>. <\/p>\n<p>It has been shown that minimizing the pignistic risk <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dae6bae3dcdac4629730754352c5e329_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#82;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> leads to choosing the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-88879811a973dafad289e3bdc75abf6e_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"18\" style=\"vertical-align: -3px;\"\/> with the highest pignistic probability [8]. If an additional rejection action <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d093ecd207b3f9d9b1c9cf0692b77d01_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"16\" style=\"vertical-align: -3px;\"\/> of constant cost <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b08f106f56be545afd0613f834360123_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#108;&#97;&#109;&#98;&#100;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"17\" style=\"vertical-align: -3px;\"\/> is possible, then the vector is rejected if <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-73d87d77e8286fb885b16a39f42eba8c_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#112;&#95;&#109;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#48;&#41;&#60;&#49;&#45;&#92;&#108;&#97;&#109;&#98;&#100;&#97;&#95;&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"127\" style=\"vertical-align: -5px;\"\/>. <\/p>\n<h3>3.4 Decision-making based on belief functions<\/h3>\n<p>Let&#8217;s now take a look at how to construct belief functions for decision-making purposes, based on the accuracy of the attributes, the accuracy of the classifiers and the output of the classification stage.<\/p>\n<p>At the credal level of the transferable belief model, in order to obtain belief functions from training data, two families of techniques are generally used: likelihood-based methods that use density estimation, and a distance-based method in which mass sets are constructed directly from distances to training vectors.<\/p>\n<ol type=\"1\">\n<li>Likelihood-based methods.<br \/>\nThe probability densities <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-58a4e06fd2df0306a3c9b56fc7dab5e4_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#70;&#40;&#120;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"60\" style=\"vertical-align: -6px;\"\/> conditional on the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/> classes are assumed to be known. Having observed <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/>, the likelihood function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1ea2d4cdda0437ba797be07de691f63b_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"58\" style=\"vertical-align: -6px;\"\/> is a <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> function in <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-18370a2fa7f2139a8d3814b85fcb03df_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#91;&#48;&#44;&#43;&#92;&#105;&#110;&#102;&#116;&#121;&#91;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"57\" style=\"vertical-align: -5px;\"\/> defined by <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-04029811d105522178b1bf92121a6acd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#76;&#40;&#119;&#95;&#113;&#124;&#120;&#41;&#61;&#102;&#40;&#120;&#124;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"142\" style=\"vertical-align: -6px;\"\/>, for all <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-74c35124316d55b1e1f6dfdbd13b55eb_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#113;&#92;&#105;&#110;&#91;&#49;&#44;&#46;&#46;&#46;&#44;&#81;&#93;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"92\" style=\"vertical-align: -5px;\"\/>. From <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-66a9f474fc3c52efdfb0ba6a70199ee8_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#76;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>, Shafer [15] proposed to construct a belief function on <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/> defined by its plausibility function as:<br \/>\n\\begin  {eqnarray}<br \/>\npl(A)=\\frac{\\max_{omega_q\\in A}[L(\\omega_q|X)]}{\\max_{omega_q\\in A}{q}[L(\\omega_q|X)]} &amp; \\forall A\\subseteq\\Omega<br \/>\n\\label{incertitude:eq:pl}\\tag{3.4.1}<br \/>\n\\end{eqnarray}Starting from axiomatic considerations, Appriou [1] proposed another method based on the construction of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2c758bec4c272382411b95fc0e7ee250_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#81;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"14\" style=\"vertical-align: -4px;\"\/> belief functions <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-496532f4919607f6aee40d808575d4a1_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#113;&#40;&#46;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"41\" style=\"vertical-align: -6px;\"\/>. The idea is to consider each class separately and evaluate the degree of belief accorded to each of them. In this case, the focal elements of each of the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d9121146a128b872d348446882fb8bdd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"23\" style=\"vertical-align: -6px;\"\/> belief functions are the singletons <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-780049cb65d8d22e60ce4b4d683234f6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"65\" style=\"vertical-align: -6px;\"\/>, the complementary subsets <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-5af0ffabd2a82533e6c21966a448c0d2_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#98;&#97;&#114;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"17\" width=\"58\" style=\"vertical-align: -6px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ffcc6ca4cbe68c6f4237bab250a62b47_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#79;&#109;&#101;&#103;&#97;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"12\" style=\"vertical-align: 0px;\"\/>. Appriou thus obtains two different models:   <\/p>\n<ul>\n<li>Model 1:<br \/>\n<a name=\"id3610831718\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 44px;\"><span class=\"ql-right-eqno\"> (3.4.2) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-efb0263d0aff9e5f9f79657d9cb2d151_l3.png\" height=\"44\" width=\"591\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#109;&#95;&#113;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#32;&#61;&#32;&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#113;&#92;&#102;&#114;&#97;&#99;&#123;&#82;&#46;&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;&#125;&#123;&#49;&#43;&#82;&#46;&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;&#125; &#109;&#95;&#113;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#32;&#61;&#32;&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#113;&#92;&#102;&#114;&#97;&#99;&#123;&#49;&#125;&#123;&#49;&#43;&#82;&#46;&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;&#125; &#109;&#95;&#113;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#32;&#61;&#32;&#49;&#45;&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#113; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<\/li>\n<li>Model 2:<br \/>\n<a name=\"id1034617641\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 19px;\"><span class=\"ql-right-eqno\"> (3.4.3) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a252d2a354b8d118b944230bb6bffd14_l3.png\" height=\"19\" width=\"612\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#109;&#95;&#113;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#32;&#61;&#32;&#48;&#92; &#109;&#95;&#113;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#32;&#61;&#32;&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#113;&#40;&#49;&#45;&#82;&#46;&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;&#41;&#92; &#109;&#95;&#113;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#32;&#61;&#32;&#49;&#45;&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#113;&#40;&#49;&#45;&#82;&#46;&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<\/li>\n<\/ul>\n<p>Here, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c537761f130c1172a67756f02b55b25d_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/> is a coefficient that can be used to model additional information (such as sensor reliability), and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dae6bae3dcdac4629730754352c5e329_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#82;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> is a normalization constant that is chosen in <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-59da6517126b2539ccc3ca241a513597_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#93;&#48;&#44;&#40;&#92;&#115;&#117;&#112;&#95;&#120;&#32;&#92;&#109;&#97;&#120;&#95;&#113;&#40;&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;&#41;&#41;&#94;&#123;&#45;&#49;&#125;&#93;\" title=\"Rendered by QuickLaTeX.com\" height=\"21\" width=\"209\" style=\"vertical-align: -6px;\"\/>. In practice, the performance of these two models appears to be equivalent. However, Appriou recommends the use of model 2, which has the advantage of being consistent with the &#8220;generalized Bayes theorem&#8221; proposed by Smets in [16]. From these <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2c758bec4c272382411b95fc0e7ee250_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#81;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"14\" style=\"vertical-align: -4px;\"\/> belief functions and using Dempster&#8217;s combination rule, a unique belief function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6b41df788161942c6f98604d37de8098_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"15\" style=\"vertical-align: 0px;\"\/> is obtained by <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1756528cf3dd282c30c543bc3247ebaa_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#61;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#113;&#109;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"18\" width=\"134\" style=\"vertical-align: -6px;\"\/>.   <\/li>\n<li>Distance-based method.<br \/>\nThe second family of models is based on distance information. An example of this is the extension of the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> nearest neighbor algorithm introduced by T. Den\u0153ux in [18]. In this method, a belief function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d6d703ffbc595d34de6fd65fdbaa27e6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"11\" width=\"21\" style=\"vertical-align: -3px;\"\/> is directly constructed using the information provided by the vectors <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c65ae606c9093b55e8ad9cced5cbf695_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;&#94;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"15\" style=\"vertical-align: 0px;\"\/> located in the neighborhood of the unknown vector <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/> by:<br \/>\n\\begin  {eqnarray}<br \/>\nm^i(\\omega_k\\})=alpha^iphi^i(d^i)\\\\<br \/>\nm^i(\\Omega) = 1-alpha^iphi^i(d^i)\\<br \/>\nm^i(A) = 0,\\forall A\\in 2^\\Omega \\backslash \\omega_k\\},\\Omega\\}<br \/>\n\\label{incertitude:eq:denoeux_modele}\\tag{3.4.4}<br \/>\n\\end{eqnarray}<br \/>\nwhere <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-7583567372cd47aeb7a55f730a69c340_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#100;&#94;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"14\" style=\"vertical-align: 0px;\"\/> is the Euclidean distance from the unknown vector to the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c65ae606c9093b55e8ad9cced5cbf695_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;&#94;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"15\" style=\"vertical-align: 0px;\"\/> vector, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-0c546cdc22558db7e2ee9a580d58f010_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#97;&#108;&#112;&#104;&#97;&#94;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"16\" style=\"vertical-align: 0px;\"\/> a parameter associated with the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-695d9d59bd04859c6c99e7feb11daab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"6\" style=\"vertical-align: 0px;\"\/>-th neighbor and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-f61fa4d7da45ce394afeb16f9b2dd481_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#112;&#104;&#105;&#94;&#105;&#40;&#100;&#94;&#105;&#41;&#61;&#101;&#94;&#123;&#45;&#92;&#103;&#97;&#109;&#109;&#97;&#94;&#105;&#40;&#100;&#94;&#105;&#41;&#94;&#50;&#125;\" title=\"Rendered by QuickLaTeX.com\" height=\"24\" width=\"129\" style=\"vertical-align: -5px;\"\/> with <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-e6b8368fbcf900943dfff704cf02aead_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#103;&#97;&#109;&#109;&#97;&#94;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"15\" style=\"vertical-align: -4px;\"\/> a positive parameter (recall that the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d5c47c3488e950cf0e59e41d4cf08832_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#98;&#97;&#99;&#107;&#115;&#108;&#97;&#115;&#104;\" title=\"Rendered by QuickLaTeX.com\" height=\"19\" width=\"8\" style=\"vertical-align: -5px;\"\/> operator means &#8220;deprived of&#8221;). The <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> nearest neighbor method yields <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-3422b6bb5c160593658b7c39425d9880_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#107;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> belief functions to be aggregated by the combination rule for decision making. <\/li>\n<\/ol>\n<h2>4 Proposal<\/h2>\n<h3>4.1 Principle<\/h3>\n<p>The theory of evidence presents several elements of interest in our quest for a measure of the accuracy associated with the decision, as well as in the integration of such a measure into this decision-making. The first of these is certainly that of being able to propose a pessimistic and optimistic boundary for the decision, which can in the first instance be brought closer to the tolerance desired in association with the decision. The second is the introduction of the &#8220;weakening&#8221; factor. This factor takes into account the reliability of the operator providing the information. This reliability, or trust, is deduced from a learning process carried out beforehand, i.e. off-line. However, as the approach is constructed, whatever the value established, whatever the object from which the measurement is established, it will be considered at the same level of importance as all those coming from the operator.     <\/p>\n<p>This should not be seen as a criticism of the construction of the attenuation factor, which was introduced to allow the combination of different information sources, such as sensors or frequency bands in MRI. Behind this proposal lies the assumption of linearity in the behavior of the information source in the face of noise and\/or the information to be acquired. While this assumption is certainly valid in the context in which this theory was developed, it is not valid in our more generic context.  <\/p>\n<p>There are two possible solutions for developing our proposal: the first would be to integrate a second factor into the formulation, which would depend solely on the object to be measured, via the operator <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-43c82d5bb00a7568d935a12e3bd969dd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"9\" style=\"vertical-align: -4px;\"\/>. The second solution is to retain a single attenuation factor integrating the two aspects to be taken into account: <\/p>\n<ul>\n<li>The contribution of the type of information considered to the decision-making process. This corresponds to <strong>a priori relevance<\/strong>, i.e. uniquely linked to the operator and considered following learning before the actual decision-making phase (the so-called on-line phase). <\/li>\n<li>The quality of the measurement, which we call precision. This precision measurement is directly dependent on the object to be measured as seen through the operator&#8217;s eyes, and is carried out to take into account non-linearities in operator or sensor behavior (e.g. responses according to different tissues in medical imaging). This precision measurement will therefore be referred to as <strong>a posteriori precision measurement<\/strong>, as it is established at the end of the measurement.  <\/li>\n<\/ul>\n<p>In both cases, we propose to retain Appriou&#8217;s notation. The combination of the various attenuation elements is written in multiplicative form. For reasons of time, we have not combined these two parts of the total accuracy estimation (a priori and a posteriori) in this work. Another reason for this decision is the problem of estimating a priori accuracy, which is highly dependent on the construction of the learning base.   <\/p>\n<p>Let&#8217;s take a look at some of the avenues we&#8217;ve explored to answer this question in the context of classification. However, as the literature on the subject is already extensive, we felt that there were already many bases for answering this question. <\/p>\n<h3>4.2 Installation<\/h3>\n<p>The belief functions per class\/classifier\/attribute are combined (Dempster) to construct a belief function per class. The decision is then made according to equation 3.3.3. The decision mechanism is illustrated in figure 4.2.1. This figure shows the sequence of processes and the exchanges between them. Each variable in the formalism is represented here, along with its origin and destination. The source of the flow illustrating the individual to be analyzed (<img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c674b8a19afef366cde8b4327cf616a6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;&#94;&#42;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"16\" style=\"vertical-align: 0px;\"\/>) is not shown, as we are not dealing with this part. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-c674b8a19afef366cde8b4327cf616a6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;&#94;&#42;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"16\" style=\"vertical-align: 0px;\"\/> can be an object injected directly into the process, but it can also be the result of subsequent processing.      <\/p>\n<figure id=\"attachment_3330\" style=\"width: 900px; margin-left: auto; margin-right: auto;\"><img loading=\"lazy\" decoding=\"async\" class=\"size-medium wp-image-3330\" src=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-300x261.png\" alt=\"DFD_chain\" width=\"900\" height=\"783\" srcset=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-300x261.png 300w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-1024x891.png 1024w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-768x668.png 768w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-1536x1336.png 1536w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-2048x1781.png 2048w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/DFD_chaine-1568x1364.png 1568w\" sizes=\"(max-width: 900px) 100vw, 900px\" \/><figcaption>Figure 4.2.1 &#8211; Data flow diagram of the processing chain<\/figcaption><\/figure>\n<p>The exploitation of Appriou&#8217;s expression, and our modified version of Appriou&#8217;s expression, exploits discrete labels based on the likelihood function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1ea2d4cdda0437ba797be07de691f63b_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"58\" style=\"vertical-align: -6px;\"\/> linking an individual <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/> to its membership of the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/>.<\/p>\n<p>In order to obtain these functions, and according to the diagram in figure 4.2.2, these labels are in our case derived from a classification stage linking the individual to the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/> class. Classically in this framework, the likelihood function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1ea2d4cdda0437ba797be07de691f63b_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"58\" style=\"vertical-align: -6px;\"\/> is then linked to the distance <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8714e6886a54fe44f39c3c81da579511_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#67;&#95;&#105;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"65\" style=\"vertical-align: -6px;\"\/> of the individual <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/> to the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6cdde80414a58b169591ab7d031aa595_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"14\" width=\"18\" style=\"vertical-align: -6px;\"\/> according to the classifier <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-695d9d59bd04859c6c99e7feb11daab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"6\" style=\"vertical-align: 0px;\"\/> (depending on the classifier, the distance can be replaced by a degree of membership or a probability). However, the ability to associate an individual with a label by a classifier is dependent on the attribute set used. Similarly, each classifier proposes a different partitioning of the attribute space, depending on the class model used or the metrics implemented. We will therefore relate the likelihood function <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-1ea2d4cdda0437ba797be07de691f63b_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#76;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#120;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"58\" style=\"vertical-align: -6px;\"\/> to the distance <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9fb1e17908f3b3c98d6f7cd5997ba61c_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#67;&#95;&#105;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#65;&#95;&#106;&#40;&#120;&#41;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"99\" style=\"vertical-align: -6px;\"\/>, where <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-27c7f19e7af285cea6d1cd185e121717_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#65;&#95;&#106;&#40;&#120;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"43\" style=\"vertical-align: -6px;\"\/> represents the vector of attributes provided by the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-43c82d5bb00a7568d935a12e3bd969dd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"9\" style=\"vertical-align: -4px;\"\/> operator (Zernike moments of order 15, for example, or a 32-bin color histogram obtained by fuzzy C-means, etc.) and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b53d1c850e3d39670961022481dc5260_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#67;&#95;&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"15\" width=\"18\" style=\"vertical-align: -3px;\"\/> refers to the <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-695d9d59bd04859c6c99e7feb11daab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"6\" style=\"vertical-align: 0px;\"\/> classifier.    <\/p>\n<figure style=\"width: 900px; margin-left: auto; margin-right: auto;\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-medium wp-image-2846\" src=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/a-diagram-of-a-process-description-automatically-300x147.png\" alt=\"A diagram of a process Description automatically generated\" width=\"900\" height=\"441\" srcset=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/a-diagram-of-a-process-description-automatically-300x147.png 300w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/a-diagram-of-a-process-description-automatically-1024x503.png 1024w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/a-diagram-of-a-process-description-automatically-768x377.png 768w, https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/02\/a-diagram-of-a-process-description-automatically.png 1428w\" sizes=\"(max-width: 900px) 100vw, 900px\" \/><figcaption>Figure 4.2.2 &#8211; Processing chain integrating attribute precision<\/figcaption><\/figure>\n<p>The choice of this script allows us to combine several labels from different classifiers and\/or different attributes. Depending on the chosen protocol, the different classifiers will either compete directly with each other, or a dynamic selection of the best classifier will take place before the merging step. In the latter case, the competition merges the labels from different attributes seen through the best classifier for each of them. The expression we arrive at is directly deduced from equation 3.3.3 :   <\/p>\n<p><a name=\"id1173485124\"><\/a><\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 65px;\"><span class=\"ql-right-eqno\"> (4.2.1) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-6fddc31f6532da50c55e65630c5cf771_l3.png\" height=\"65\" width=\"670\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#92;&#108;&#101;&#102;&#116;&#92;&#123; &#92;&#98;&#101;&#103;&#105;&#110;&#123;&#97;&#114;&#114;&#97;&#121;&#125;&#123;&#108;&#108;&#125; &#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#92;&#125;&#41;&#38;&#61;&#48;&#92;&#92;&igrave; &#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#38;&#61;&#92;&#97;&#108;&#112;&#104;&#97;&#95;&#123;&#106;&#113;&#125;&#40;&#49;&#45;&#82;&#95;&#123;&#105;&#106;&#113;&#125;&#46;&#67;&#95;&#105;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#65;&#95;&#106;&#40;&#120;&#41;&#41;&#41;&#92;&#92;&#92; &#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#38;&#61;&#49;&#45;&#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#92;&#92; &#92;&#101;&#110;&#100;&#123;&#97;&#114;&#114;&#97;&#121;&#125; &#92;&#114;&#105;&#103;&#104;&#116;&#46;&#92; &#92;&#110;&#111;&#110;&#117;&#109;&#98;&#101;&#114;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#123;&#105;&#125;&#92;&#105;&#110;&#91;&#49;&#44;&#110;&#95;&#99;&#93;&#44;&#32;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#123;&#106;&#125;&#92;&#105;&#110;&#91;&#49;&#44;&#110;&#95;&#97;&#93;&#44;&#32;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#123;&#113;&#125;&#92;&#105;&#110;&#91;&#49;&#44;&#110;&#95;&#113;&#93; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>Appriou incorporates <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dae6bae3dcdac4629730754352c5e329_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#82;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/>, a normalization parameter of the formula established so that the belief mass <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-2179f178e2cb42c1f6965aacc13b96d6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"65\" style=\"vertical-align: -6px;\"\/> is between 0 and 1. <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dae6bae3dcdac4629730754352c5e329_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#82;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> is therefore a value between <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a5e437be25f29374d30f66cd46adf81c_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#48;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"9\" style=\"vertical-align: 0px;\"\/> and the maximum value that can be estimated for <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-35e08f9b99fa4a05a7e67ab208f89608_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#67;&#95;&#105;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#124;&#65;&#95;&#105;&#40;&#120;&#41;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"97\" style=\"vertical-align: -6px;\"\/>. This value of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-dae6bae3dcdac4629730754352c5e329_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#82;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"14\" style=\"vertical-align: 0px;\"\/> is unique for classifier <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-695d9d59bd04859c6c99e7feb11daab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"6\" style=\"vertical-align: 0px;\"\/> and corresponds to the greatest distance identified or possible between an individual and the classes defined for this classifier, hence:<br \/>\n\\begin  {eqnarray}<br \/>\n\\nonumber\\mbox{for }0\\leq 1-\\frac{x}{y}\\leq 1\\mbox{ it is necessary }y\\geq x\\<br \/>\n\\nonumber\\mbox{ hence }y=sup_x(\\max_q(C_i(\\omega_q|A_j(x))))\\<br \/>\n0&lt;R\\leq\\frac{1}{\\sup_x(\\max_q(C_i(\\omega_q|A_j(x))))}<br \/>\n\\tag{4.2.2}<br \/>\n\\end{eqnarray}<\/p>\n<p>The belief mass we&#8217;ve just established is <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-af03dba7556ed72f91758ea5f39cfb92_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#92;&#125;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"123\" style=\"vertical-align: -6px;\"\/>, i.e. the belief mass associated with the fact that class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ac7da57d7f507262338bb5168feb3e06_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"8\" style=\"vertical-align: -4px;\"\/> is assigned to individual <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/> by classifier <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-695d9d59bd04859c6c99e7feb11daab6_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#105;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"6\" style=\"vertical-align: 0px;\"\/> for attribute vector <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-43c82d5bb00a7568d935a12e3bd969dd_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#106;\" title=\"Rendered by QuickLaTeX.com\" height=\"16\" width=\"9\" style=\"vertical-align: -4px;\"\/>. What we&#8217;re interested in is the final decision to assign the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ac7da57d7f507262338bb5168feb3e06_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"8\" style=\"vertical-align: -4px;\"\/> to the individual <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ede05c264bba0eda080918aaa09c4658_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#120;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"10\" style=\"vertical-align: 0px;\"\/>, and consequently the mass of belief in the class <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ac7da57d7f507262338bb5168feb3e06_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"8\" style=\"vertical-align: -4px;\"\/>, <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-8c5e0b90c59678b0b501507dfd4ff9a5_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#113;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"103\" style=\"vertical-align: -6px;\"\/>, and the associated masses <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-43542a3610aa5793112beba21856582a_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#113;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"54\" style=\"vertical-align: -6px;\"\/> and <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-369020661825bfbacc5060d8eb689430_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#109;&#95;&#113;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;\" title=\"Rendered by QuickLaTeX.com\" height=\"20\" width=\"49\" style=\"vertical-align: -6px;\"\/> : <\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 43px;\"><span class=\"ql-right-eqno\"> (4.2.3) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-94273d979c36e500c792414eda76d46a_l3.png\" height=\"43\" width=\"227\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#92;&#108;&#101;&#102;&#116;&#92;&#123; &#92;&#98;&#101;&#103;&#105;&#110;&#123;&#97;&#114;&#114;&#97;&#121;&#125;&#123;&#108;&#125; &#109;&#95;&#113;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#92;&#125;&#41;&#92;&#92; &#109;&#95;&#113;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#113;&#125;&#41;&#92; &#109;&#95;&#113;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41; &#92;&#101;&#110;&#100;&#123;&#97;&#114;&#114;&#97;&#121;&#125; &#92;&#114;&#105;&#103;&#104;&#116;&#46;&#92; &#92;&#110;&#111;&#110;&#117;&#109;&#98;&#101;&#114;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#113;&#92;&#105;&#110;&#91;&#49;&#44;&#110;&#93; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>Based on Dempster&#8217;s combination model, we formalize this decision making from <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-ac7da57d7f507262338bb5168feb3e06_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#113;\" title=\"Rendered by QuickLaTeX.com\" height=\"12\" width=\"8\" style=\"vertical-align: -4px;\"\/> intermediate decisions on a set of <img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-b170995d512c659d8668b4e42e1fef6b_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#110;\" title=\"Rendered by QuickLaTeX.com\" height=\"8\" width=\"11\" style=\"vertical-align: 0px;\"\/> labels for a single set of attributes:<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 47px;\"><span class=\"ql-right-eqno\"> (4.2.4) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9dc80e0964c881f79e0aafe5c78e2859_l3.png\" height=\"47\" width=\"586\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125; &#92;&#108;&#101;&#102;&#116;&#92;&#123; &#92;&#98;&#101;&#103;&#105;&#110;&#123;&#97;&#114;&#114;&#97;&#121;&#125;&#123;&#108;&#108;&#125; &#109;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#125;&#41;&#38;&#61;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#113;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#113;&#125;&#109;&#95;&#123;&#105;&#113;&#125;&#40;&#92;&#123;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#125;&#41;&#92; &#109;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#125;&#41;&#38;&#61;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#113;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#113;&#125;&#109;&#95;&#123;&#105;&#113;&#125;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#125;&#41; &#92;&#101;&#110;&#100;&#123;&#97;&#114;&#114;&#97;&#121;&#125; &#92;&#114;&#105;&#103;&#104;&#116;&#46;&#92; &#92;&#110;&#111;&#110;&#117;&#109;&#98;&#101;&#114;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#113;&#92;&#105;&#110;&#91;&#49;&#44;&#110;&#93;&#44;&#32;&#92;&#102;&#111;&#114;&#97;&#108;&#108;&#32;&#105;&#92;&#105;&#110;&#91;&#49;&#44;&#110;&#95;&#99;&#93; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#110;&#97;&#114;&#114;&#97;&#121;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>and :<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 49px;\"><span class=\"ql-right-eqno\"> (4.2.5) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-a952d35f44e0bba28376802af47982ad_l3.png\" height=\"49\" width=\"257\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#109;&#40;&#92;&#79;&#109;&#101;&#103;&#97;&#41;&#61;&#49;&#45;&#92;&#115;&#117;&#109;&#95;&#123;&#105;&#61;&#49;&#125;&#94;&#123;&#110;&#125;&#40;&#109;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#105;&#92;&#125;&#41;&#43;&#109;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#105;&#125;&#41;&#41; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>More generally, in the presence of several sets of attributes from different operators (or different parameters), Dempster&#8217;s combination rule leads to the general formulation :<\/p>\n<p class=\"ql-center-displayed-equation\" style=\"line-height: 47px;\"><span class=\"ql-right-eqno\"> (4.2.6) <\/span><span class=\"ql-left-eqno\"> &nbsp; <\/span><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-d0a7335a004e5ed53ba8d1dd01514855_l3.png\" height=\"47\" width=\"545\" class=\"ql-img-displayed-equation quicklatex-auto-format\" alt=\"&#92;&#98;&#101;&#103;&#105;&#110;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125; &#92;&#108;&#101;&#102;&#116;&#92;&#123; &#92;&#98;&#101;&#103;&#105;&#110;&#123;&#97;&#114;&#114;&#97;&#121;&#125;&#123;&#108;&#108;&#125; &#109;&#40;&#92;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#125;&#41;&#38;&#61;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#113;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#113;&#125;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#105;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#99;&#125;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#106;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#97;&#125;&#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#92;&#125;&#41; &#109;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#125;&#41;&#38;&#61;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#113;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#113;&#125;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#105;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#99;&#125;&#92;&#98;&#105;&#103;&#111;&#112;&#108;&#117;&#115;&#95;&#123;&#106;&#61;&#49;&#125;&#94;&#123;&#110;&#95;&#97;&#125;&#109;&#95;&#123;&#105;&#106;&#113;&#125;&#40;&#92;&#111;&#118;&#101;&#114;&#108;&#105;&#110;&#101;&#123;&#92;&#111;&#109;&#101;&#103;&#97;&#95;&#107;&#125;&#41;&#92;&#92; &#92;&#101;&#110;&#100;&#123;&#97;&#114;&#114;&#97;&#121;&#125; &#92;&#114;&#105;&#103;&#104;&#116;&#46; &#92;&#101;&#110;&#100;&#123;&#101;&#113;&#117;&#97;&#116;&#105;&#111;&#110;&#42;&#125;\" title=\"Rendered by QuickLaTeX.com\"\/><\/p>\n<p>The data flow diagram in figure 4.2.1 illustrates how this script works.<\/p>\n<h2>Bibliography<\/h2>\n<p>[1] A. Appriou. Probabilities and uncertainties in multi-sensor data fusion. <i>Revue Scientifique et Technique de la D\u00e9fense<\/i>, 11 :27-40, 1991.<br \/>\n[2] D. Arrivault. <i>Apport des graphes dans la reconnaissance non-contrainte de caract\u00e8res manuscrits anciens<\/i>. PhD thesis, Universit\u00e9 de Poitiers, 2002.<br \/>\n[3] I. Bloch. Fusion d&#8217;informations num\u00e9riques: panorama m\u00e9thodologique. In <i>Journ\u00e9es Nationales de la Recherche en Robotique<\/i>, 2005.[7]<br \/>\n[4] R. Caruana and A. Niculescu-Mizil. An empirical comparison of supervised learning algorithms. In <i>ICML &#8217;06: Proceedings of the 23rd international conference on Machine learning<\/i>, pages 161-168, New York, NY, USA, 2006. ACM.<br \/>\n[5] T. M. Cover and P. E. Hart. Nearest neighbor pattern classification.          <i>IEEE Trans. on Information Theory <\/i>13(1):21-27, 1967.<br \/>\n[6] A. P. Dempster. Upper and lower probabilities induced by multiple valued mappings. <i>Annals of Mathematical Statistics<\/i>, 38 :325-339, 1967.<br \/>\n[7] A. P. Dempster. A generalization of bayesian inference. <i>Journal of the Royal Statistical Society<\/i>, 30 :205-247, 1968.<br \/>\n[8] T. Den\u0153ux. Analysis of evidence-theoretic decision rules for pattern classification. <i>Pattern Recognition<\/i>, 30(7) :1095-1107, 1997.<br \/>\n[9] J. Kholas and P. A. Monney. <i>Lecture Notes in Economics and Mathematical Systems<\/i>, volume 425, chapter A mathematical theory of hints: An approach to the Dempster-Shafer theory of evidence. Springer-Verlag, 1995.<br \/>\n[10] G. J. Klir and M. J. Wierman.      <i>Uncertainty-based information. Elements of generalized information theory, 2nd edition. Studies in fuzzyness and soft computing  <\/i>. Physica-Verlag, 1999.<br \/>\n[11] E. Lefevre, O. Colot, and P. Vannoorenberghe. Belief function combination and conflict management. <i>Information Fusion<\/i>, 3 :149-162, 2002.<br \/>\n[12] D. Mercier, B. Quost, and T. Deno\u0153ux. Refined modeling of sensor reliability in the belief function framework using contextual discounting. <i>Information Fusion<\/i>, 2007.<br \/>\n[13] E. Ramasso. <i>Recognition of state sequences by the Transferable Belief Model &#8211; Application to the analysis of athletics videos<\/i>. PhD thesis, Universit\u00e9 Joseph Fourier de Grenoble, 2007.<br \/>\n[14] I. Rish. An empirical study of the naive bayes classifier. In <i>IJCAI-01 workshop on &#8220;Empirical Methods in AI&#8221;<\/i>, 2001.<br \/>\n[15] G. Shafer. <i>A Mathematical Theory of Evidence<\/i>. Princeton University Press, 1976.<br \/>\n[16] P. Smets. Beliefs functions: The disjunctive rule of combination and the generalized bayesian theorem.         <i>Int. Jour. of Approximate Reasoning  <\/i>9 :1-35, 1993.<br \/>\n[17] P. Smets. The nature of the unnormalized beliefs encountered in the transferable belief model. In <i>Proceedings of the 8th Annual Conference on Uncertainty in Artificial Intelligence<\/i>, pages 292-297, 1992.<br \/>\n[18] P. Smets. <i>Advances in the Dempster-Shafer Theory of Evidence &#8211; What is Dempster-Shafer&#8217;s model?<\/i> Wiley, 1994.<br \/>\n[19] P. Smets. Decision making in the tbm: The necessity of the pignistic transformation.     <i>Int. Jour. of Approximate Reasoning  <\/i>38 :133-147, 2005.<br \/>\n[20] P. Smets and R. Kennes. The transferable belief model. <i>Artificial Intelligence<\/i>, 66(2) :191-234, 1994.<br \/>\n[21] B. Yaghlane, P. Smets, and K. Mellouli. Yaghlane, P. Smets, and K. Mellouli. Belief function independence: I. the marginal case.    <i>Int. Jour. of Approximate Reasoning  <\/i>29 :47-70, 2001.<br \/>\n[22] B. Yaghlane, P. Smets, and K. Mellouli. Belief function independence : II. the conditionnal case.    <i>Int. Jour. of Approximate Reasoning  <\/i>31(1) :31-75, 2002.<br \/>\n[23] B. Yaghlane, P. Smets, and K. Mellouli. Independence concept for belief functions. In Physica-Verlag, editor, <i>Technologies for constructing intelligent systems : Tools<\/i>, 2002 [93].  <\/p>\n<p id=\"footnote-1\" style=\"color: black; font-weight: normal;\"><img decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-262bcc0158d697c58f339ad8fa642a53_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#94;&#49;\" title=\"Rendered by QuickLaTeX.com\" height=\"9\" width=\"5\" style=\"vertical-align: 6px;\"\/> <a href=\"https:\/\/eur-lex.europa.eu\/legal-content\/FR\/TXT\/?uri=CELEX%3A52021PC0206\">Proposal for a Regulation of the European Parliament and of the Council laying down harmonized rules on artificial intelligence (artificial intelligence legislation) and amending certain Union legislative acts<\/a><\/p>\n<p id=\"footnote-2\" style=\"color: black; font-weight: normal;\"><img decoding=\"async\" src=\"https:\/\/www.a2d.ai\/wp-content\/ql-cache\/quicklatex.com-9129dfc8cd44dea5a511583766cf1b07_l3.png\" class=\"ql-img-inline-formula quicklatex-auto-format\" alt=\"&#94;&#50;\" title=\"Rendered by QuickLaTeX.com\" height=\"9\" width=\"7\" style=\"vertical-align: 6px;\"\/> <a href=\"https:\/\/www.iso.org\/standard\/81230.html\">ISO\/IEC 42001<\/a><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>How to decide in the presence of imprecision? S. Sochacki, PhD in Computer Science A2D, ZA Les Minimes, 1 rue de la Trinquette, 17000 La Rochelle, FRANCE 1 Introduction Imprecision\/uncertainty\/conflict AI Act, explainability How do you decide in IT? What does an automatic system base its decisions on? Very often, this kind of system is [&hellip;]<\/p>\n","protected":false},"author":5,"featured_media":3359,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[33],"tags":[],"class_list":["post-5704","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology"],"acf":[],"yoast_head":"<title>Deciding in the presence of imprecision | A2D<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Deciding in the presence of imprecision | A2D\" \/>\n<meta property=\"og:description\" content=\"How to decide in the presence of imprecision? S. Sochacki, PhD in Computer Science A2D, ZA Les Minimes, 1 rue de la Trinquette, 17000 La Rochelle, FRANCE 1 Introduction Imprecision\/uncertainty\/conflict AI Act, explainability How do you decide in IT? What does an automatic system base its decisions on? Very often, this kind of system is [&hellip;]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/\" \/>\n<meta property=\"og:site_name\" content=\"A2D\" \/>\n<meta property=\"article:published_time\" content=\"2024-03-13T08:28:31+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/03\/istockphoto-1776536866-612x612-1.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"612\" \/>\n\t<meta property=\"og:image:height\" content=\"408\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Stephane SOCHACKI\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Stephane SOCHACKI\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"41 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/\"},\"author\":{\"name\":\"Stephane SOCHACKI\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/#\\\/schema\\\/person\\\/a0558201d87ec3e252e6e278470bf324\"},\"headline\":\"Deciding in the presence of imprecision\",\"datePublished\":\"2024-03-13T08:28:31+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/\"},\"wordCount\":8293,\"image\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.a2d.ai\\\/wp-content\\\/uploads\\\/2024\\\/03\\\/istockphoto-1776536866-612x612-1.jpg\",\"articleSection\":[\"Technology\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/\",\"url\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/\",\"name\":\"Deciding in the presence of imprecision | A2D\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.a2d.ai\\\/wp-content\\\/uploads\\\/2024\\\/03\\\/istockphoto-1776536866-612x612-1.jpg\",\"datePublished\":\"2024-03-13T08:28:31+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/#\\\/schema\\\/person\\\/a0558201d87ec3e252e6e278470bf324\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#primaryimage\",\"url\":\"https:\\\/\\\/www.a2d.ai\\\/wp-content\\\/uploads\\\/2024\\\/03\\\/istockphoto-1776536866-612x612-1.jpg\",\"contentUrl\":\"https:\\\/\\\/www.a2d.ai\\\/wp-content\\\/uploads\\\/2024\\\/03\\\/istockphoto-1776536866-612x612-1.jpg\",\"width\":612,\"height\":408,\"caption\":\"A conceptual image showcasing the merging of advanced robotics and human interaction against an urban skyline. A robotic arm and a human hand point towards a glowing holographic question mark, symbolizing the future of technology, questions, and potential challenges.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/deciding-in-the-presence-of-imprecision\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/home-2\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Deciding in the presence of imprecision\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/#website\",\"url\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/\",\"name\":\"A2D\",\"description\":\"Acquire to Decide\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.a2d.ai\\\/en\\\/#\\\/schema\\\/person\\\/a0558201d87ec3e252e6e278470bf324\",\"name\":\"Stephane SOCHACKI\"}]}<\/script>","yoast_head_json":{"title":"Deciding in the presence of imprecision | A2D","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/","og_locale":"en_US","og_type":"article","og_title":"Deciding in the presence of imprecision | A2D","og_description":"How to decide in the presence of imprecision? S. Sochacki, PhD in Computer Science A2D, ZA Les Minimes, 1 rue de la Trinquette, 17000 La Rochelle, FRANCE 1 Introduction Imprecision\/uncertainty\/conflict AI Act, explainability How do you decide in IT? What does an automatic system base its decisions on? Very often, this kind of system is [&hellip;]","og_url":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/","og_site_name":"A2D","article_published_time":"2024-03-13T08:28:31+00:00","og_image":[{"width":612,"height":408,"url":"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/03\/istockphoto-1776536866-612x612-1.jpg","type":"image\/jpeg"}],"author":"Stephane SOCHACKI","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Stephane SOCHACKI","Est. reading time":"41 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#article","isPartOf":{"@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/"},"author":{"name":"Stephane SOCHACKI","@id":"https:\/\/www.a2d.ai\/en\/#\/schema\/person\/a0558201d87ec3e252e6e278470bf324"},"headline":"Deciding in the presence of imprecision","datePublished":"2024-03-13T08:28:31+00:00","mainEntityOfPage":{"@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/"},"wordCount":8293,"image":{"@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#primaryimage"},"thumbnailUrl":"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/03\/istockphoto-1776536866-612x612-1.jpg","articleSection":["Technology"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/","url":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/","name":"Deciding in the presence of imprecision | A2D","isPartOf":{"@id":"https:\/\/www.a2d.ai\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#primaryimage"},"image":{"@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#primaryimage"},"thumbnailUrl":"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/03\/istockphoto-1776536866-612x612-1.jpg","datePublished":"2024-03-13T08:28:31+00:00","author":{"@id":"https:\/\/www.a2d.ai\/en\/#\/schema\/person\/a0558201d87ec3e252e6e278470bf324"},"breadcrumb":{"@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#primaryimage","url":"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/03\/istockphoto-1776536866-612x612-1.jpg","contentUrl":"https:\/\/www.a2d.ai\/wp-content\/uploads\/2024\/03\/istockphoto-1776536866-612x612-1.jpg","width":612,"height":408,"caption":"A conceptual image showcasing the merging of advanced robotics and human interaction against an urban skyline. A robotic arm and a human hand point towards a glowing holographic question mark, symbolizing the future of technology, questions, and potential challenges."},{"@type":"BreadcrumbList","@id":"https:\/\/www.a2d.ai\/en\/deciding-in-the-presence-of-imprecision\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.a2d.ai\/en\/home-2\/"},{"@type":"ListItem","position":2,"name":"Deciding in the presence of imprecision"}]},{"@type":"WebSite","@id":"https:\/\/www.a2d.ai\/en\/#website","url":"https:\/\/www.a2d.ai\/en\/","name":"A2D","description":"Acquire to Decide","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.a2d.ai\/en\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/www.a2d.ai\/en\/#\/schema\/person\/a0558201d87ec3e252e6e278470bf324","name":"Stephane SOCHACKI"}]}},"_links":{"self":[{"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/posts\/5704","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/comments?post=5704"}],"version-history":[{"count":0,"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/posts\/5704\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/media\/3359"}],"wp:attachment":[{"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/media?parent=5704"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/categories?post=5704"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.a2d.ai\/en\/wp-json\/wp\/v2\/tags?post=5704"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}