{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:QPX2RUUIMTOAJMOVLX2MWZDAVL","short_pith_number":"pith:QPX2RUUI","schema_version":"1.0","canonical_sha256":"83efa8d28864dc04b1d55df4cb6460aadd1bdba16386aa1564820842bc97636f","source":{"kind":"arxiv","id":"2202.01017","version":2},"attestation_state":"computed","paper":{"title":"Multi-Task Learning as a Bargaining Game","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.GT"],"primary_cat":"cs.LG","authors_text":"Aviv Navon, Aviv Shamsian, Ethan Fetaya, Gal Chechik, Haggai Maron, Idan Achituve, Kenji Kawaguchi","submitted_at":"2022-02-02T13:21:53Z","abstract_excerpt":"In Multi-task learning (MTL), a joint model is trained to simultaneously make predictions for several tasks. Joint training reduces computation costs and improves data efficiency; however, since the gradients of these different tasks may conflict, training a joint model for MTL often yields lower performance than its corresponding single-task counterparts. A common method for alleviating this issue is to combine per-task gradients into a joint update direction using a particular heuristic. In this paper, we propose viewing the gradients combination step as a bargaining game, where tasks negoti"},"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":"2202.01017","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-02-02T13:21:53Z","cross_cats_sorted":["cs.GT"],"title_canon_sha256":"02e736df9b2a08afa23d057650026a78946006205b4d7d9b1ce37d18c8d90f7d","abstract_canon_sha256":"cae33ee67fce820ccbf6f1fb9bebfb7d2e71c41c759d3957f62908fed5682f23"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:38:34.973923Z","signature_b64":"yAlp3KKHQDjpWvs3Vjmfs7gsjrDErXPlT4JQRlljRca/x6CPzlvnVireR/dhQ/5QL+9uMyLfbxLqqnZ2BIj3Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"83efa8d28864dc04b1d55df4cb6460aadd1bdba16386aa1564820842bc97636f","last_reissued_at":"2026-07-05T04:38:34.973515Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:38:34.973515Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-Task Learning as a Bargaining Game","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.GT"],"primary_cat":"cs.LG","authors_text":"Aviv Navon, Aviv Shamsian, Ethan Fetaya, Gal Chechik, Haggai Maron, Idan Achituve, Kenji Kawaguchi","submitted_at":"2022-02-02T13:21:53Z","abstract_excerpt":"In Multi-task learning (MTL), a joint model is trained to simultaneously make predictions for several tasks. Joint training reduces computation costs and improves data efficiency; however, since the gradients of these different tasks may conflict, training a joint model for MTL often yields lower performance than its corresponding single-task counterparts. A common method for alleviating this issue is to combine per-task gradients into a joint update direction using a particular heuristic. In this paper, we propose viewing the gradients combination step as a bargaining game, where tasks negoti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.01017","kind":"arxiv","version":2},"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/2202.01017/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":"2202.01017","created_at":"2026-07-05T04:38:34.973573+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.01017v2","created_at":"2026-07-05T04:38:34.973573+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.01017","created_at":"2026-07-05T04:38:34.973573+00:00"},{"alias_kind":"pith_short_12","alias_value":"QPX2RUUIMTOA","created_at":"2026-07-05T04:38:34.973573+00:00"},{"alias_kind":"pith_short_16","alias_value":"QPX2RUUIMTOAJMOV","created_at":"2026-07-05T04:38:34.973573+00:00"},{"alias_kind":"pith_short_8","alias_value":"QPX2RUUI","created_at":"2026-07-05T04:38:34.973573+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27377","citing_title":"DanceOPD: On-Policy Generative Field Distillation","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17493","citing_title":"When Robots Sleep: Offline Skill Consolidation for Shared-Policy Robot Learning","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00293","citing_title":"Rosetta: Composable Native Multimodal Pretraining","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2510.11491","citing_title":"Constraint-Aware Reinforcement Learning via Adaptive Action Scaling","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14350","citing_title":"Distributionally Robust Multi-Task Reinforcement Learning via Adaptive Task Sampling","ref_index":223,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08939","citing_title":"Delve into the Applicability of Advanced Optimizers for Multi-Task Learning","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL","json":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL.json","graph_json":"https://pith.science/api/pith-number/QPX2RUUIMTOAJMOVLX2MWZDAVL/graph.json","events_json":"https://pith.science/api/pith-number/QPX2RUUIMTOAJMOVLX2MWZDAVL/events.json","paper":"https://pith.science/paper/QPX2RUUI"},"agent_actions":{"view_html":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL","download_json":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL.json","view_paper":"https://pith.science/paper/QPX2RUUI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.01017&json=true","fetch_graph":"https://pith.science/api/pith-number/QPX2RUUIMTOAJMOVLX2MWZDAVL/graph.json","fetch_events":"https://pith.science/api/pith-number/QPX2RUUIMTOAJMOVLX2MWZDAVL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL/action/storage_attestation","attest_author":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL/action/author_attestation","sign_citation":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL/action/citation_signature","submit_replication":"https://pith.science/pith/QPX2RUUIMTOAJMOVLX2MWZDAVL/action/replication_record"}},"created_at":"2026-07-05T04:38:34.973573+00:00","updated_at":"2026-07-05T04:38:34.973573+00:00"}