<?xml version="1.0" encoding="utf-8"?>
<TEI xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:hal="http://hal.archives-ouvertes.fr/" xmlns:gml="http://www.opengis.net/gml/3.3/" xmlns:gmlce="http://www.opengis.net/gml/3.3/ce" version="1.1" xsi:schemaLocation="http://www.tei-c.org/ns/1.0 http://api.archives-ouvertes.fr/documents/aofr-sword.xsd">
  <teiHeader>
    <fileDesc>
      <titleStmt>
        <title>HAL TEI export of hal-03996206</title>
      </titleStmt>
      <publicationStmt>
        <distributor>CCSD</distributor>
        <availability status="restricted">
          <licence target="https://creativecommons.org/publicdomain/zero/1.0/">CC0 1.0 - Universal</licence>
        </availability>
        <date when="2026-05-01T23:39:41+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">Why is the prediction wrong? Towards underfitting case explanation via meta-classification</title>
            <author role="aut">
              <persName>
                <forename type="first">Sheng</forename>
                <surname>Zhou</surname>
              </persName>
              <idno type="halauthorid">864917-0</idno>
              <affiliation ref="#struct-553459"/>
              <affiliation ref="#struct-560488"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Pierre</forename>
                <surname>Blanchart</surname>
              </persName>
              <email type="md5">e961796c2fac0cfacb5246f24b9b461a</email>
              <email type="domain">gmail.com</email>
              <idno type="idhal" notation="numeric">1228858</idno>
              <idno type="halauthorid" notation="string">610811-1228858</idno>
              <idno type="IDREF">https://www.idref.fr/187538913</idno>
              <idno type="VIAF">https://viaf.org/viaf/317199779</idno>
              <affiliation ref="#struct-560488"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Michel</forename>
                <surname>Crucianu</surname>
              </persName>
              <email type="md5">0c715a026c697c14721a3ef8d61ba3aa</email>
              <email type="domain">cnam.fr</email>
              <idno type="idhal" notation="string">michel-crucianu</idno>
              <idno type="idhal" notation="numeric">180351</idno>
              <idno type="halauthorid" notation="string">2139-180351</idno>
              <idno type="ORCID">https://orcid.org/0000-0001-8204-6843</idno>
              <affiliation ref="#struct-553459"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Marin</forename>
                <surname>Ferecatu</surname>
              </persName>
              <email type="md5">8e918f983b5002a2ffa6b5d9d4146e03</email>
              <email type="domain">lecnam.net</email>
              <idno type="idhal" notation="string">ferecatm</idno>
              <idno type="idhal" notation="numeric">180193</idno>
              <idno type="halauthorid" notation="string">24786-180193</idno>
              <idno type="ORCID">https://orcid.org/0000-0003-3845-1758</idno>
              <idno type="IDREF">https://www.idref.fr/099256126</idno>
              <affiliation ref="#struct-553459"/>
            </author>
            <editor role="depositor">
              <persName>
                <forename>Marin</forename>
                <surname>FERECATU</surname>
              </persName>
              <email type="md5">8e918f983b5002a2ffa6b5d9d4146e03</email>
              <email type="domain">lecnam.net</email>
            </editor>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2023-02-19 16:28:10</date>
              <date type="whenModified">2025-07-09 15:06:18</date>
              <date type="whenReleased">2023-02-20 13:41:19</date>
              <date type="whenProduced">2022-10-13</date>
              <date type="whenEndEmbargoed">2023-02-19</date>
              <ref type="file" target="https://cnam.hal.science/hal-03996206v1/document">
                <date notBefore="2023-02-19"/>
              </ref>
              <ref type="file" subtype="author" n="1" target="https://cnam.hal.science/hal-03996206v1/file/paper.pdf" id="file-3996206-3484287">
                <date notBefore="2023-02-19"/>
              </ref>
              <ref type="externalLink" target="http://arxiv.org/pdf/2302.09952"/>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="819997">
                <persName>
                  <forename>Marin</forename>
                  <surname>FERECATU</surname>
                </persName>
                <email type="md5">8e918f983b5002a2ffa6b5d9d4146e03</email>
                <email type="domain">lecnam.net</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-03996206</idno>
            <idno type="halUri">https://cnam.hal.science/hal-03996206</idno>
            <idno type="halBibtex">zhou:hal-03996206</idno>
            <idno type="halRefHtml">&lt;i&gt;2022 IEEE 9th International Conference on Data Science and Advanced Analytics (DSAA)&lt;/i&gt;, Oct 2022, Shenzhen, China. pp.10032332, &lt;a target="_blank" href="https://dx.doi.org/10.1109/DSAA54385.2022.10032332"&gt;&amp;#x27E8;10.1109/DSAA54385.2022.10032332&amp;#x27E9;&lt;/a&gt;</idno>
            <idno type="halRef">2022 IEEE 9th International Conference on Data Science and Advanced Analytics (DSAA), Oct 2022, Shenzhen, China. pp.10032332, &amp;#x27E8;10.1109/DSAA54385.2022.10032332&amp;#x27E9;</idno>
            <availability status="restricted">
              <licence target="https://about.hal.science/hal-authorisation-v1/">HAL Authorization<ref corresp="#file-3996206-3484287"/></licence>
            </availability>
          </publicationStmt>
          <seriesStmt>
            <idno type="stamp" n="CEA">CEA - Commissariat à l'énergie atomique</idno>
            <idno type="stamp" n="CNAM">Conservatoire National des Arts et Métiers</idno>
            <idno type="stamp" n="DRT" corresp="CEA">Direction de la recherche technologique</idno>
            <idno type="stamp" n="CEA-UPSAY" corresp="CEA">CEA - Université Paris-Saclay</idno>
            <idno type="stamp" n="UNIV-PARIS-SACLAY">Université Paris-Saclay</idno>
            <idno type="stamp" n="LIST" corresp="CEA">Laboratoire d'Intégration des Systèmes et des Technologies</idno>
            <idno type="stamp" n="CEDRIC-CNAM" corresp="CNAM">Centre d'études et de recherche en informatique et communications </idno>
            <idno type="stamp" n="DM2I" corresp="LIST">Département Métrologie Instrumentation &amp; Information</idno>
            <idno type="stamp" n="UNIVERSITE-PARIS-SACLAY" corresp="UNIV-PARIS-SACLAY">Université Paris-Saclay</idno>
            <idno type="stamp" n="GS-COMPUTER-SCIENCE">Graduate School Computer Science</idno>
            <idno type="stamp" n="GS-SPORT-HUMAN-MOVEMENT">Graduate School Sport, Mouvement, Facteurs Humains</idno>
            <idno type="stamp" n="TEST3-HALCNRS">TEST3-HALCNRS</idno>
            <idno type="stamp" n="HESAM-CNAM" corresp="HESAM">Conservatoire National des Arts et Métiers - Hesam</idno>
            <idno type="stamp" n="HESAM">Hesam</idno>
            <idno type="stamp" n="DIN" corresp="DRT">Département d'Instrumentation Numérique</idno>
          </seriesStmt>
          <notesStmt>
            <note type="audience" n="2">International</note>
            <note type="invited" n="0">No</note>
            <note type="popular" n="0">No</note>
            <note type="peer" n="1">Yes</note>
            <note type="proceedings" n="1">Yes</note>
          </notesStmt>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">Why is the prediction wrong? Towards underfitting case explanation via meta-classification</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Sheng</forename>
                    <surname>Zhou</surname>
                  </persName>
                  <idno type="halauthorid">864917-0</idno>
                  <affiliation ref="#struct-553459"/>
                  <affiliation ref="#struct-560488"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Pierre</forename>
                    <surname>Blanchart</surname>
                  </persName>
                  <email type="md5">e961796c2fac0cfacb5246f24b9b461a</email>
                  <email type="domain">gmail.com</email>
                  <idno type="idhal" notation="numeric">1228858</idno>
                  <idno type="halauthorid" notation="string">610811-1228858</idno>
                  <idno type="IDREF">https://www.idref.fr/187538913</idno>
                  <idno type="VIAF">https://viaf.org/viaf/317199779</idno>
                  <affiliation ref="#struct-560488"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Michel</forename>
                    <surname>Crucianu</surname>
                  </persName>
                  <email type="md5">0c715a026c697c14721a3ef8d61ba3aa</email>
                  <email type="domain">cnam.fr</email>
                  <idno type="idhal" notation="string">michel-crucianu</idno>
                  <idno type="idhal" notation="numeric">180351</idno>
                  <idno type="halauthorid" notation="string">2139-180351</idno>
                  <idno type="ORCID">https://orcid.org/0000-0001-8204-6843</idno>
                  <affiliation ref="#struct-553459"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Marin</forename>
                    <surname>Ferecatu</surname>
                  </persName>
                  <email type="md5">8e918f983b5002a2ffa6b5d9d4146e03</email>
                  <email type="domain">lecnam.net</email>
                  <idno type="idhal" notation="string">ferecatm</idno>
                  <idno type="idhal" notation="numeric">180193</idno>
                  <idno type="halauthorid" notation="string">24786-180193</idno>
                  <idno type="ORCID">https://orcid.org/0000-0003-3845-1758</idno>
                  <idno type="IDREF">https://www.idref.fr/099256126</idno>
                  <affiliation ref="#struct-553459"/>
                </author>
              </analytic>
              <monogr>
                <meeting>
                  <title>2022 IEEE 9th International Conference on Data Science and Advanced Analytics (DSAA)</title>
                  <date type="start">2022-10-13</date>
                  <date type="end">2022-10-16</date>
                  <settlement>Shenzhen</settlement>
                  <country key="CN">China</country>
                </meeting>
                <imprint>
                  <publisher>IEEE</publisher>
                  <biblScope unit="pp">10032332</biblScope>
                  <date type="datePub">2023-02-19</date>
                </imprint>
              </monogr>
              <idno type="arxiv">2302.09952</idno>
              <idno type="doi">10.1109/DSAA54385.2022.10032332</idno>
              <ref type="publisher">https://ieeexplore.ieee.org/abstract/document/10032332/</ref>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="en">English</language>
            </langUsage>
            <textClass>
              <keywords scheme="author">
                <term xml:lang="en">kNN</term>
                <term xml:lang="en">Online learning</term>
                <term xml:lang="en">Trustworthy Artificial intelligence</term>
                <term xml:lang="en">accuracy</term>
                <term xml:lang="en">diagnosis</term>
                <term xml:lang="en">decision tree</term>
                <term xml:lang="en">miss-classification</term>
                <term xml:lang="en">faulty data</term>
                <term xml:lang="en">classification</term>
                <term xml:lang="en">Artificial Intelligence</term>
                <term xml:lang="en">Model debugging</term>
                <term xml:lang="en">Model validation</term>
                <term xml:lang="en">MLP</term>
                <term xml:lang="en">XGBoost</term>
                <term xml:lang="en">Explainability</term>
                <term xml:lang="en">Interpretability</term>
                <term xml:lang="en">Machine learning</term>
              </keywords>
              <classCode scheme="halDomain" n="info.info-lg">Computer Science [cs]/Machine Learning [cs.LG]</classCode>
              <classCode scheme="halDomain" n="info.info-ai">Computer Science [cs]/Artificial Intelligence [cs.AI]</classCode>
              <classCode scheme="halTypology" n="COMM">Conference papers</classCode>
              <classCode scheme="halOldTypology" n="COMM">Conference papers</classCode>
              <classCode scheme="halTreeTypology" n="COMM">Conference papers</classCode>
            </textClass>
            <abstract xml:lang="en">
              <p>In this paper we present a heuristic method to provide individual explanations for those elements in a dataset (data points) which are wrongly predicted by a given classifier. Since the general case is too difficult, in the present work we focus on faulty data from an underfitted model. First, we project the faulty data into a hand-crafted, and thus human readable, intermediate representation (meta-representation, profile vectors), with the aim of separating the two main causes of miss-classification: the classifier is not strong enough, or the data point belongs to an area of the input space where classes are not separable. Second, in the space of these profile vectors, we present a method to fit a meta-classifier (decision tree) and express its output as a set of interpretable (human readable) explanation rules, which leads to several target diagnosis labels: data point is either correctly classified, or faulty due to a too weak model, or faulty due to mixed (overlapped) classes in the input space. Experimental results on several real datasets show more than 80% diagnosis label accuracy and confirm that the proposed intermediate representation allows to achieve a high degree of invariance with respect to the classifier used in the input space and to the dataset being classified, i.e. we can learn the metaclassifier on a dataset with a given classifier and successfully predict diagnosis labels for a different dataset or classifier (or both).</p>
            </abstract>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="researchteam" xml:id="struct-553459" status="VALID">
          <orgName>CEDRIC. Données complexes, apprentissage et représentations</orgName>
          <orgName type="acronym">CEDRIC - VERTIGO</orgName>
          <desc>
            <address>
              <country key="FR"/>
            </address>
            <ref type="url">https://cedric.cnam.fr/index.php/labo/Vertigo</ref>
          </desc>
          <listRelation>
            <relation active="#struct-16574" type="direct"/>
            <relation name="EA4629" active="#struct-179741" type="indirect"/>
            <relation name="EA4629" active="#struct-300351" type="indirect"/>
          </listRelation>
        </org>
        <org type="researchteam" xml:id="struct-560488" status="OLD">
          <orgName>Laboratoire Sciences des Données et de la Décision (CEA, LIST)</orgName>
          <orgName type="acronym">LS2D (CEA, LIST)</orgName>
          <date type="start">2019-01-01</date>
          <date type="end">2022-12-31</date>
          <desc>
            <address>
              <addrLine>anciennement Laboratoire Analyse des Données et Intelligence des Systèmes (LADIS), Laboratoire Outils d'Analyse des Données (LOAD), Laboratoire des Processus Stochastiques et Stationnaires (LPSS)</addrLine>
              <country key="FR"/>
            </address>
          </desc>
          <listRelation>
            <relation name="DRT/LIST/DM2I/SID/LS2D" active="#struct-528161" type="direct"/>
            <relation name="DRT/LIST/DM2I" active="#struct-40217" type="indirect"/>
            <relation name="DRT/LIST" active="#struct-440043" type="indirect"/>
            <relation name="DRT" active="#struct-300016" type="indirect"/>
            <relation active="#struct-419361" type="indirect"/>
          </listRelation>
        </org>
        <org type="laboratory" xml:id="struct-16574" status="VALID">
          <idno type="IdRef">155463381</idno>
          <idno type="RNSR">201119480F</idno>
          <idno type="ROR">https://ror.org/044j5mm75</idno>
          <orgName>Centre d'études et de recherche en informatique et communications</orgName>
          <orgName type="acronym">CEDRIC</orgName>
          <desc>
            <address>
              <addrLine>292 rue Saint-Martin 75141 PARIS Cedex 03Case courrier 2D4P30</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://cedric.cnam.fr/</ref>
          </desc>
          <listRelation>
            <relation name="EA4629" active="#struct-179741" type="direct"/>
            <relation name="EA4629" active="#struct-300351" type="direct"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-179741" status="VALID">
          <orgName>Ecole Nationale Supérieure d'Informatique pour l'Industrie et l'Entreprise</orgName>
          <orgName type="acronym">ENSIIE</orgName>
          <desc>
            <address>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.ensiie.fr/</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-300351" status="VALID">
          <idno type="IdRef">027404978</idno>
          <idno type="ISNI">0000000121690459</idno>
          <idno type="ROR">https://ror.org/0175hh227</idno>
          <orgName>Conservatoire National des Arts et Métiers [Cnam]</orgName>
          <orgName type="acronym">Cnam</orgName>
          <desc>
            <address>
              <addrLine>292 Rue Saint-Martin, 75003 Paris</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.cnam.fr/</ref>
          </desc>
        </org>
        <org type="department" xml:id="struct-528161" status="OLD">
          <idno type="RNSR">200822621H</idno>
          <orgName>Département Métrologie Instrumentation &amp; Information (CEA, LIST)</orgName>
          <orgName type="acronym">DM2I (CEA, LIST)</orgName>
          <date type="start">2013-04-01</date>
          <date type="end">2022-12-31</date>
          <desc>
            <address>
              <addrLine>ex DCSI.ex DETECS.CEA LIST</addrLine>
              <country key="FR"/>
            </address>
          </desc>
          <listRelation>
            <relation name="DRT/LIST/DM2I" active="#struct-40217" type="direct"/>
            <relation name="DRT/LIST" active="#struct-440043" type="indirect"/>
            <relation name="DRT" active="#struct-300016" type="indirect"/>
            <relation active="#struct-419361" type="direct"/>
          </listRelation>
        </org>
        <org type="laboratory" xml:id="struct-40217" status="VALID">
          <idno type="IdRef">156836882</idno>
          <idno type="ISNI">0000 0004 0405 1788</idno>
          <idno type="RNSR">200118591H</idno>
          <idno type="ROR">https://ror.org/000dbcc61</idno>
          <idno type="Wikidata">Q30299467</idno>
          <orgName>Laboratoire d'Intégration des Systèmes et des Technologies</orgName>
          <orgName type="acronym">LIST (CEA)</orgName>
          <desc>
            <address>
              <addrLine>DRT/LISTNano-INNOVAvenue de la Vauve91120 Palaiseau</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www-list.cea.fr/</ref>
          </desc>
          <listRelation>
            <relation name="DRT/LIST" active="#struct-440043" type="direct"/>
            <relation name="DRT" active="#struct-300016" type="indirect"/>
          </listRelation>
        </org>
        <org type="regrouplaboratory" xml:id="struct-440043" status="VALID">
          <idno type="IdRef">067087930</idno>
          <idno type="ISNI">0000000121157881</idno>
          <idno type="RNSR">199018589D</idno>
          <idno type="ROR">https://ror.org/02ggzyd20</idno>
          <idno type="Wikidata">Q30299418</idno>
          <orgName>Direction de Recherche Technologique (CEA)</orgName>
          <orgName type="acronym">DRT (CEA)</orgName>
          <desc>
            <address>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.drt-cea.com/</ref>
          </desc>
          <listRelation>
            <relation name="DRT" active="#struct-300016" type="direct"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-300016" status="VALID">
          <idno type="IdRef">026372061</idno>
          <idno type="ISNI">0000000122998025</idno>
          <idno type="ROR">https://ror.org/00jjx8s55</idno>
          <idno type="Wikidata">Q868550</idno>
          <orgName>Commissariat à l'énergie atomique et aux énergies alternatives</orgName>
          <orgName type="acronym">CEA</orgName>
          <desc>
            <address>
              <addrLine>Centre de SaclayCentre de GrenobleCentre de Cadaracheetc</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.cea.fr/</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-419361" status="VALID">
          <idno type="IdRef">241345251</idno>
          <idno type="ROR">https://ror.org/03xjwb503</idno>
          <orgName>Université Paris-Saclay</orgName>
          <desc>
            <address>
              <addrLine>Bâtiment Bréguet, 3 Rue Joliot Curie 2e ét, 91190 Gif-sur-Yvette</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.universite-paris-saclay.fr/fr</ref>
          </desc>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>