{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2QO7JJTQBHCQX423DTTZEYRLYR","short_pith_number":"pith:2QO7JJTQ","schema_version":"1.0","canonical_sha256":"d41df4a67009c50bf35b1ce792622bc45c95dccc0faf3c5fbbbb0e255093a297","source":{"kind":"arxiv","id":"2506.17592","version":1},"attestation_state":"computed","paper":{"title":"SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Changick Kim, Minsuk Jang, Myung-Joon Kwon, Wonjun Lee, Younghun Kim","submitted_at":"2025-06-21T05:11:35Z","abstract_excerpt":"Face identity provides a powerful signal for deepfake detection. Prior studies show that even when not explicitly modeled, classifiers often learn identity features implicitly. This has led to conflicting views: some suppress identity cues to reduce bias, while others rely on them as forensic evidence. To reconcile these views, we analyze two hypotheses: (1) whether face identity alone is discriminative for detecting deepfakes, and (2) whether such identity features generalize poorly across manipulation methods. Our experiments confirm that identity is informative but context-dependent. While "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2506.17592","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-06-21T05:11:35Z","cross_cats_sorted":[],"title_canon_sha256":"c9c6b3cdb09b29c314019b3e88ca8d173f7a7dc3f3ec62cc03156d96a9d3daad","abstract_canon_sha256":"ac4b86687a844b50f9f71e0b260f97cc39d4269be9f1a73b2dbb1892eaffb105"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:25:20.012372Z","signature_b64":"O98omigatiAY/VY7FmyMdFqrvsx83MlYlQ3hP5SdX3rqmDa1kPCHXxN0iiKWs6gf07XRTvpPTyA0Lyi5ZX6uCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d41df4a67009c50bf35b1ce792622bc45c95dccc0faf3c5fbbbb0e255093a297","last_reissued_at":"2026-07-05T11:25:20.011884Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:25:20.011884Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Changick Kim, Minsuk Jang, Myung-Joon Kwon, Wonjun Lee, Younghun Kim","submitted_at":"2025-06-21T05:11:35Z","abstract_excerpt":"Face identity provides a powerful signal for deepfake detection. Prior studies show that even when not explicitly modeled, classifiers often learn identity features implicitly. This has led to conflicting views: some suppress identity cues to reduce bias, while others rely on them as forensic evidence. To reconcile these views, we analyze two hypotheses: (1) whether face identity alone is discriminative for detecting deepfakes, and (2) whether such identity features generalize poorly across manipulation methods. Our experiments confirm that identity is informative but context-dependent. While "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.17592","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.17592/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2506.17592","created_at":"2026-07-05T11:25:20.011945+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.17592v1","created_at":"2026-07-05T11:25:20.011945+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.17592","created_at":"2026-07-05T11:25:20.011945+00:00"},{"alias_kind":"pith_short_12","alias_value":"2QO7JJTQBHCQ","created_at":"2026-07-05T11:25:20.011945+00:00"},{"alias_kind":"pith_short_16","alias_value":"2QO7JJTQBHCQX423","created_at":"2026-07-05T11:25:20.011945+00:00"},{"alias_kind":"pith_short_8","alias_value":"2QO7JJTQ","created_at":"2026-07-05T11:25:20.011945+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.18238","citing_title":"Non-Colliding Biometric Identities for Digital Entities: Geometry, Capacity, and Million-Scale Virtual Identity Provisioning","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR","json":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR.json","graph_json":"https://pith.science/api/pith-number/2QO7JJTQBHCQX423DTTZEYRLYR/graph.json","events_json":"https://pith.science/api/pith-number/2QO7JJTQBHCQX423DTTZEYRLYR/events.json","paper":"https://pith.science/paper/2QO7JJTQ"},"agent_actions":{"view_html":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR","download_json":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR.json","view_paper":"https://pith.science/paper/2QO7JJTQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.17592&json=true","fetch_graph":"https://pith.science/api/pith-number/2QO7JJTQBHCQX423DTTZEYRLYR/graph.json","fetch_events":"https://pith.science/api/pith-number/2QO7JJTQBHCQX423DTTZEYRLYR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR/action/storage_attestation","attest_author":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR/action/author_attestation","sign_citation":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR/action/citation_signature","submit_replication":"https://pith.science/pith/2QO7JJTQBHCQX423DTTZEYRLYR/action/replication_record"}},"created_at":"2026-07-05T11:25:20.011945+00:00","updated_at":"2026-07-05T11:25:20.011945+00:00"}