{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GRPMHXN6GNVGC3CPOCP52EGTOL","short_pith_number":"pith:GRPMHXN6","schema_version":"1.0","canonical_sha256":"345ec3ddbe336a616c4f709fdd10d372f79cce973817d1a7b661d4169ae5343e","source":{"kind":"arxiv","id":"2407.15580","version":3},"attestation_state":"computed","paper":{"title":"Annealed Multiple Choice Learning: Overcoming limitations of Winner-takes-all with annealing","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.SD","eess.AS","math.PR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Adrien Cort\\'es, David Perera, Ga\\\"el Richard, Mickael Chen, Slim Essid, Th\\'eo Mariotte, Victor Letzelter","submitted_at":"2024-07-22T12:16:56Z","abstract_excerpt":"We introduce Annealed Multiple Choice Learning (aMCL) which combines simulated annealing with MCL. MCL is a learning framework handling ambiguous tasks by predicting a small set of plausible hypotheses. These hypotheses are trained using the Winner-takes-all (WTA) scheme, which promotes the diversity of the predictions. However, this scheme may converge toward an arbitrarily suboptimal local minimum, due to the greedy nature of WTA. We overcome this limitation using annealing, which enhances the exploration of the hypothesis space during training. We leverage insights from statistical physics "},"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":"2407.15580","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-22T12:16:56Z","cross_cats_sorted":["cs.SD","eess.AS","math.PR","stat.ML"],"title_canon_sha256":"dfd72fb853d1f50172ea4a8a01ecc215f2973d83bc04acd93238c168c5d97c90","abstract_canon_sha256":"eb312d50e2ab402e69df50f567ec4fb9eeed39f45dd654e27d8c2c5bf6d4deed"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:02:05.250852Z","signature_b64":"jHKuSEW7nytSfb8zvT+c1T2gA2rqF55bh6FrOAHqU4xlllGuaRfN2xFMbEke+hqVRz+EAxWa1mtWcyI4MRoFBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"345ec3ddbe336a616c4f709fdd10d372f79cce973817d1a7b661d4169ae5343e","last_reissued_at":"2026-07-05T10:02:05.250296Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:02:05.250296Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Annealed Multiple Choice Learning: Overcoming limitations of Winner-takes-all with annealing","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.SD","eess.AS","math.PR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Adrien Cort\\'es, David Perera, Ga\\\"el Richard, Mickael Chen, Slim Essid, Th\\'eo Mariotte, Victor Letzelter","submitted_at":"2024-07-22T12:16:56Z","abstract_excerpt":"We introduce Annealed Multiple Choice Learning (aMCL) which combines simulated annealing with MCL. MCL is a learning framework handling ambiguous tasks by predicting a small set of plausible hypotheses. These hypotheses are trained using the Winner-takes-all (WTA) scheme, which promotes the diversity of the predictions. However, this scheme may converge toward an arbitrarily suboptimal local minimum, due to the greedy nature of WTA. We overcome this limitation using annealing, which enhances the exploration of the hypothesis space during training. We leverage insights from statistical physics "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.15580","kind":"arxiv","version":3},"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/2407.15580/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":"2407.15580","created_at":"2026-07-05T10:02:05.250361+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.15580v3","created_at":"2026-07-05T10:02:05.250361+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.15580","created_at":"2026-07-05T10:02:05.250361+00:00"},{"alias_kind":"pith_short_12","alias_value":"GRPMHXN6GNVG","created_at":"2026-07-05T10:02:05.250361+00:00"},{"alias_kind":"pith_short_16","alias_value":"GRPMHXN6GNVGC3CP","created_at":"2026-07-05T10:02:05.250361+00:00"},{"alias_kind":"pith_short_8","alias_value":"GRPMHXN6","created_at":"2026-07-05T10:02:05.250361+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.18497","citing_title":"Multiple Choice Learning for Efficient Speech Separation with Many Speakers","ref_index":23,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL","json":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL.json","graph_json":"https://pith.science/api/pith-number/GRPMHXN6GNVGC3CPOCP52EGTOL/graph.json","events_json":"https://pith.science/api/pith-number/GRPMHXN6GNVGC3CPOCP52EGTOL/events.json","paper":"https://pith.science/paper/GRPMHXN6"},"agent_actions":{"view_html":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL","download_json":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL.json","view_paper":"https://pith.science/paper/GRPMHXN6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.15580&json=true","fetch_graph":"https://pith.science/api/pith-number/GRPMHXN6GNVGC3CPOCP52EGTOL/graph.json","fetch_events":"https://pith.science/api/pith-number/GRPMHXN6GNVGC3CPOCP52EGTOL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL/action/storage_attestation","attest_author":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL/action/author_attestation","sign_citation":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL/action/citation_signature","submit_replication":"https://pith.science/pith/GRPMHXN6GNVGC3CPOCP52EGTOL/action/replication_record"}},"created_at":"2026-07-05T10:02:05.250361+00:00","updated_at":"2026-07-05T10:02:05.250361+00:00"}