<?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-05016500</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="2026-05-15T23:50:28+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">How much is the Source Mismatch an Important Problem for Deepfake Detection ? *</title>
            <author role="aut">
              <persName>
                <forename type="first">Antoine</forename>
                <surname>Mallet</surname>
              </persName>
              <email type="md5">3c2d24d0871ba272c29befedd71c89c6</email>
              <email type="domain">utt.fr</email>
              <idno type="idhal" notation="string">antoinemallet</idno>
              <idno type="idhal" notation="numeric">1381424</idno>
              <idno type="halauthorid" notation="string">2049437-1381424</idno>
              <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=P-ACEIAAAAAJ</idno>
              <idno type="ORCID">https://orcid.org/0009-0006-0479-564X</idno>
              <affiliation ref="#struct-93484"/>
            </author>
            <author role="crp">
              <persName>
                <forename type="first">Rémi</forename>
                <surname>Cogranne</surname>
              </persName>
              <email type="md5">c4e497271b4bcc69e75434f225de3df3</email>
              <email type="domain">utt.fr</email>
              <idno type="idhal" notation="string">remi-cogranne</idno>
              <idno type="idhal" notation="numeric">9636</idno>
              <idno type="halauthorid" notation="string">13565-9636</idno>
              <idno type="ORCID">https://orcid.org/0000-0002-4205-4694</idno>
              <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=ZKpouPAAAAAJ</idno>
              <idno type="IDREF">https://www.idref.fr/168828960</idno>
              <affiliation ref="#struct-93484"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Minoru</forename>
                <surname>Kuribayashi</surname>
              </persName>
              <idno type="halauthorid">3368261-0</idno>
              <affiliation ref="#struct-311564"/>
              <affiliation ref="#struct-311564"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Arthur</forename>
                <surname>Méreur</surname>
              </persName>
              <email type="md5">841e58c2f5c774f0be4c45496a72c609</email>
              <email type="domain">utt.fr</email>
              <idno type="idhal" notation="numeric">1389294</idno>
              <idno type="halauthorid" notation="string">3166873-1389294</idno>
              <affiliation ref="#struct-93484"/>
            </author>
            <editor role="depositor">
              <persName>
                <forename>Rémi</forename>
                <surname>Cogranne</surname>
              </persName>
              <email type="md5">c4e497271b4bcc69e75434f225de3df3</email>
              <email type="domain">utt.fr</email>
            </editor>
            <funder ref="#projanr-55961"/>
            <funder ref="#projanr-125051"/>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2025-04-02 04:44:48</date>
              <date type="whenModified">2026-01-15 15:52:03</date>
              <date type="whenReleased">2025-04-11 12:07:42</date>
              <date type="whenProduced">2025-04-02</date>
              <date type="whenEndEmbargoed">2025-04-02</date>
              <ref type="file" target="https://hal.science/hal-05016500v1/document">
                <date notBefore="2025-04-02"/>
              </ref>
              <ref type="file" subtype="author" n="1" target="https://hal.science/hal-05016500v1/file/Deepfake_final_HAL.pdf" id="file-5016500-4346095">
                <date notBefore="2025-04-02"/>
              </ref>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="164186">
                <persName>
                  <forename>Rémi</forename>
                  <surname>Cogranne</surname>
                </persName>
                <email type="md5">c4e497271b4bcc69e75434f225de3df3</email>
                <email type="domain">utt.fr</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-05016500</idno>
            <idno type="halUri">https://hal.science/hal-05016500</idno>
            <idno type="halBibtex">mallet:hal-05016500</idno>
            <idno type="halRefHtml">&lt;i&gt;APSIPA Transactions on Signal and Information Processing&lt;/i&gt;, In press, pp.18</idno>
            <idno type="halRef">APSIPA Transactions on Signal and Information Processing, In press, pp.18</idno>
            <availability status="restricted">
              <licence target="https://creativecommons.org/licenses/by/4.0/">CC BY 4.0 - Attribution<ref corresp="#file-5016500-4346095"/></licence>
            </availability>
          </publicationStmt>
          <seriesStmt>
            <idno type="stamp" n="UNIV-TROYES">Université de Technologie de Troyes</idno>
            <idno type="stamp" n="UTT">UTT</idno>
            <idno type="stamp" n="ANR">ANR</idno>
            <idno type="stamp" n="UTT-GESTION" corresp="UNIV-TROYES">UTT - Gestion interne</idno>
            <idno type="stamp" n="UTT-FULL-TEXT">UTT - Texte intégral</idno>
            <idno type="stamp" n="CYBERSCURITE">PEPR Cybersécurité</idno>
            <idno type="stamp" n="ANR-COMPROMIS">ANR-COMPROMIS</idno>
          </seriesStmt>
          <notesStmt>
            <note type="audience" n="2">International</note>
            <note type="popular" n="0">No</note>
            <note type="peer" n="1">Yes</note>
          </notesStmt>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">How much is the Source Mismatch an Important Problem for Deepfake Detection ? *</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Antoine</forename>
                    <surname>Mallet</surname>
                  </persName>
                  <email type="md5">3c2d24d0871ba272c29befedd71c89c6</email>
                  <email type="domain">utt.fr</email>
                  <idno type="idhal" notation="string">antoinemallet</idno>
                  <idno type="idhal" notation="numeric">1381424</idno>
                  <idno type="halauthorid" notation="string">2049437-1381424</idno>
                  <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=P-ACEIAAAAAJ</idno>
                  <idno type="ORCID">https://orcid.org/0009-0006-0479-564X</idno>
                  <affiliation ref="#struct-93484"/>
                </author>
                <author role="crp">
                  <persName>
                    <forename type="first">Rémi</forename>
                    <surname>Cogranne</surname>
                  </persName>
                  <email type="md5">c4e497271b4bcc69e75434f225de3df3</email>
                  <email type="domain">utt.fr</email>
                  <idno type="idhal" notation="string">remi-cogranne</idno>
                  <idno type="idhal" notation="numeric">9636</idno>
                  <idno type="halauthorid" notation="string">13565-9636</idno>
                  <idno type="ORCID">https://orcid.org/0000-0002-4205-4694</idno>
                  <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=ZKpouPAAAAAJ</idno>
                  <idno type="IDREF">https://www.idref.fr/168828960</idno>
                  <affiliation ref="#struct-93484"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Minoru</forename>
                    <surname>Kuribayashi</surname>
                  </persName>
                  <idno type="halauthorid">3368261-0</idno>
                  <affiliation ref="#struct-311564"/>
                  <affiliation ref="#struct-311564"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Arthur</forename>
                    <surname>Méreur</surname>
                  </persName>
                  <email type="md5">841e58c2f5c774f0be4c45496a72c609</email>
                  <email type="domain">utt.fr</email>
                  <idno type="idhal" notation="numeric">1389294</idno>
                  <idno type="halauthorid" notation="string">3166873-1389294</idno>
                  <affiliation ref="#struct-93484"/>
                </author>
              </analytic>
              <monogr>
                <idno type="halJournalId" status="VALID">101488</idno>
                <idno type="eissn">2048-7703</idno>
                <title level="j">APSIPA Transactions on Signal and Information Processing</title>
                <imprint>
                  <publisher>Cambridge University Press</publisher>
                  <biblScope unit="pp">18</biblScope>
                  <date type="datePub" subtype="inPress">2025-04-02</date>
                </imprint>
              </monogr>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="en">English</language>
            </langUsage>
            <textClass>
              <keywords scheme="author">
                <term xml:lang="en">Source Mismatch</term>
                <term xml:lang="en">Detection</term>
                <term xml:lang="en">Distribution-shift</term>
                <term xml:lang="en">Empirical evaluation</term>
                <term xml:lang="en">Experimental methods</term>
                <term xml:lang="en">Generative AI</term>
                <term xml:lang="en">Diffusion Model</term>
                <term xml:lang="en">Deepfake</term>
              </keywords>
              <classCode scheme="halDomain" n="info.info-ai">Computer Science [cs]/Artificial Intelligence [cs.AI]</classCode>
              <classCode scheme="halDomain" n="info.info-cr">Computer Science [cs]/Cryptography and Security [cs.CR]</classCode>
              <classCode scheme="halDomain" n="info.info-ts">Computer Science [cs]/Signal and Image Processing</classCode>
              <classCode scheme="halTypology" n="ART">Journal articles</classCode>
              <classCode scheme="halOldTypology" n="ART">Journal articles</classCode>
              <classCode scheme="halTreeTypology" n="ART">Journal articles</classCode>
            </textClass>
            <abstract xml:lang="en">
              <p>&lt;div&gt;&lt;p&gt;Over the past few decades, AI generative methods have advanced significantly, making it increasingly challenging to distinguish genuine photographs from AI-generated images, sometimes also referred to as deepfakes. In response, numerous deepfake detection methods and models have been developed, achieving high accuracy. However, the evaluation of these detection methods is often limited to a single dataset, which is typically created by generating multiple images using a specific deepfake generation methods and a fixed set of hyperparameters. This dataset is then randomly split into training and testing sets, but such an approach cannot take into account the variations of hyperparameters on deepfake detection performance. This paper addresses the fundamental question of source mismatch, where a model is trained on a specific deepfake generation source (including hyperparameters) and tested on a different one, highlighting the need to investigate the causes and impacts of such a mismatch as well as to develop solutions to this critical issue.&lt;/p&gt;&lt;/div&gt;</p>
            </abstract>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="institution" xml:id="struct-93484" status="VALID">
          <idno type="ROR">https://ror.org/01qhqcj41</idno>
          <orgName>Université de Technologie de Troyes</orgName>
          <orgName type="acronym">UTT</orgName>
          <desc>
            <address>
              <addrLine>12 rue Marie CurieCS 4206010004 TROYES CEDEX</addrLine>
              <country key="FR"/>
            </address>
          </desc>
        </org>
        <org type="regroupinstitution" xml:id="struct-311564" status="VALID">
          <idno type="ROR">https://ror.org/01dq60k83</idno>
          <orgName>Tohoku University [Sendai]</orgName>
          <desc>
            <address>
              <addrLine>2-1-1 Katahira, Aoba-ku, Sendai, 980-8577 Japan</addrLine>
              <country key="JP"/>
            </address>
            <ref type="url">https://www.tohoku.ac.jp/en/</ref>
          </desc>
        </org>
      </listOrg>
      <listOrg type="projects">
        <org type="anrProject" xml:id="projanr-55961" status="VALID">
          <idno type="anr">ANR-21-CE39-0002</idno>
          <orgName>PACeS</orgName>
          <desc>Protection contre les usages criminels de la stéganographie</desc>
          <date type="start">2021</date>
        </org>
        <org type="anrProject" xml:id="projanr-125051" status="VALID">
          <idno type="anr">ANR-22-PECY-0011</idno>
          <orgName>COMPROMIS</orgName>
          <desc>COntenus Multimedia PROtégés par Machines IntelligenteS</desc>
          <date type="start">2022</date>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>