{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FX4ZMNJAI73EDIO3PEBO4LFZJT","short_pith_number":"pith:FX4ZMNJA","schema_version":"1.0","canonical_sha256":"2df996352047f641a1db7902ee2cb94cfa7e5a4381654d5dc76fcdf7e2f418de","source":{"kind":"arxiv","id":"2503.04046","version":1},"attestation_state":"computed","paper":{"title":"Continual Optimization with Symmetry Teleportation for Multi-Task Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chunyan Miao, Peilin Zhao, Pengcheng Wu, Zhipeng Zhou, Ziqiao Meng","submitted_at":"2025-03-06T02:58:09Z","abstract_excerpt":"Multi-task learning (MTL) is a widely explored paradigm that enables the simultaneous learning of multiple tasks using a single model. Despite numerous solutions, the key issues of optimization conflict and task imbalance remain under-addressed, limiting performance. Unlike existing optimization-based approaches that typically reweight task losses or gradients to mitigate conflicts or promote progress, we propose a novel approach based on Continual Optimization with Symmetry Teleportation (COST). During MTL optimization, when an optimization conflict arises, we seek an alternative loss-equival"},"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":"2503.04046","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-06T02:58:09Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9eb79cc7dc81d1962e3c6bd2aef6334709a55c6fe4ae22f7b2df9eea4dcef0ed","abstract_canon_sha256":"f2d89e23f33e90089ee5d72b6bdad4565391d4d2cd034835cc42d779b035fb6c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:25:27.364085Z","signature_b64":"IgwLUiM4OTQEocYOeA+7TB4q+DWo+JvRJsVqFbZ1hws8povP5Yt8NiyHrferBYya5eCr+hlfIsq7W+v9rEj/Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2df996352047f641a1db7902ee2cb94cfa7e5a4381654d5dc76fcdf7e2f418de","last_reissued_at":"2026-07-05T10:25:27.363571Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:25:27.363571Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Continual Optimization with Symmetry Teleportation for Multi-Task Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chunyan Miao, Peilin Zhao, Pengcheng Wu, Zhipeng Zhou, Ziqiao Meng","submitted_at":"2025-03-06T02:58:09Z","abstract_excerpt":"Multi-task learning (MTL) is a widely explored paradigm that enables the simultaneous learning of multiple tasks using a single model. Despite numerous solutions, the key issues of optimization conflict and task imbalance remain under-addressed, limiting performance. Unlike existing optimization-based approaches that typically reweight task losses or gradients to mitigate conflicts or promote progress, we propose a novel approach based on Continual Optimization with Symmetry Teleportation (COST). During MTL optimization, when an optimization conflict arises, we seek an alternative loss-equival"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.04046","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/2503.04046/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":"2503.04046","created_at":"2026-07-05T10:25:27.363644+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.04046v1","created_at":"2026-07-05T10:25:27.363644+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.04046","created_at":"2026-07-05T10:25:27.363644+00:00"},{"alias_kind":"pith_short_12","alias_value":"FX4ZMNJAI73E","created_at":"2026-07-05T10:25:27.363644+00:00"},{"alias_kind":"pith_short_16","alias_value":"FX4ZMNJAI73EDIO3","created_at":"2026-07-05T10:25:27.363644+00:00"},{"alias_kind":"pith_short_8","alias_value":"FX4ZMNJA","created_at":"2026-07-05T10:25:27.363644+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT","json":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT.json","graph_json":"https://pith.science/api/pith-number/FX4ZMNJAI73EDIO3PEBO4LFZJT/graph.json","events_json":"https://pith.science/api/pith-number/FX4ZMNJAI73EDIO3PEBO4LFZJT/events.json","paper":"https://pith.science/paper/FX4ZMNJA"},"agent_actions":{"view_html":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT","download_json":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT.json","view_paper":"https://pith.science/paper/FX4ZMNJA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.04046&json=true","fetch_graph":"https://pith.science/api/pith-number/FX4ZMNJAI73EDIO3PEBO4LFZJT/graph.json","fetch_events":"https://pith.science/api/pith-number/FX4ZMNJAI73EDIO3PEBO4LFZJT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT/action/storage_attestation","attest_author":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT/action/author_attestation","sign_citation":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT/action/citation_signature","submit_replication":"https://pith.science/pith/FX4ZMNJAI73EDIO3PEBO4LFZJT/action/replication_record"}},"created_at":"2026-07-05T10:25:27.363644+00:00","updated_at":"2026-07-05T10:25:27.363644+00:00"}