{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DNC3O576MEYEGV2JA7LNVO2TDW","short_pith_number":"pith:DNC3O576","schema_version":"1.0","canonical_sha256":"1b45b777fe613043574907d6dabb531db4bd9694e79855d6efefec0649edce68","source":{"kind":"arxiv","id":"2502.00217","version":1},"attestation_state":"computed","paper":{"title":"Fantastic Multi-Task Gradient Updates and How to Find Them In a Cone","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Chao Gao, Chunhua Zhou, Kunlin Zhang, Muhammad Kamran Janjua, Negar Hassanpour, Sepehr Lavasani, Xiaowen Zhang","submitted_at":"2025-01-31T23:11:12Z","abstract_excerpt":"Balancing competing objectives remains a fundamental challenge in multi-task learning (MTL), primarily due to conflicting gradients across individual tasks. A common solution relies on computing a dynamic gradient update vector that balances competing tasks as optimization progresses. Building on this idea, we propose ConicGrad, a principled, scalable, and robust MTL approach formulated as a constrained optimization problem. Our method introduces an angular constraint to dynamically regulate gradient update directions, confining them within a cone centered on the reference gradient of the over"},"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":"2502.00217","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-31T23:11:12Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"0f7cfb9293c23a9d9122e9d95f95741a8af6d1f3d8c74ac3de61391b96eb3a6d","abstract_canon_sha256":"84db31a54ace0d6c7253c26ed1d99e85d7259730558a738495b29f547cf9eb20"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:08:30.666663Z","signature_b64":"nylVtvu0uiQ/6bY28f2uz356Fygm5j0/6emVcpznxdO86f/d9GXR1BjLao0Sax5bnELRCax3OaytLg6kDGnABw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1b45b777fe613043574907d6dabb531db4bd9694e79855d6efefec0649edce68","last_reissued_at":"2026-07-05T10:08:30.666174Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:08:30.666174Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fantastic Multi-Task Gradient Updates and How to Find Them In a Cone","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Chao Gao, Chunhua Zhou, Kunlin Zhang, Muhammad Kamran Janjua, Negar Hassanpour, Sepehr Lavasani, Xiaowen Zhang","submitted_at":"2025-01-31T23:11:12Z","abstract_excerpt":"Balancing competing objectives remains a fundamental challenge in multi-task learning (MTL), primarily due to conflicting gradients across individual tasks. A common solution relies on computing a dynamic gradient update vector that balances competing tasks as optimization progresses. Building on this idea, we propose ConicGrad, a principled, scalable, and robust MTL approach formulated as a constrained optimization problem. Our method introduces an angular constraint to dynamically regulate gradient update directions, confining them within a cone centered on the reference gradient of the over"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.00217","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/2502.00217/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":"2502.00217","created_at":"2026-07-05T10:08:30.666236+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.00217v1","created_at":"2026-07-05T10:08:30.666236+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.00217","created_at":"2026-07-05T10:08:30.666236+00:00"},{"alias_kind":"pith_short_12","alias_value":"DNC3O576MEYE","created_at":"2026-07-05T10:08:30.666236+00:00"},{"alias_kind":"pith_short_16","alias_value":"DNC3O576MEYEGV2J","created_at":"2026-07-05T10:08:30.666236+00:00"},{"alias_kind":"pith_short_8","alias_value":"DNC3O576","created_at":"2026-07-05T10:08:30.666236+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/DNC3O576MEYEGV2JA7LNVO2TDW","json":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW.json","graph_json":"https://pith.science/api/pith-number/DNC3O576MEYEGV2JA7LNVO2TDW/graph.json","events_json":"https://pith.science/api/pith-number/DNC3O576MEYEGV2JA7LNVO2TDW/events.json","paper":"https://pith.science/paper/DNC3O576"},"agent_actions":{"view_html":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW","download_json":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW.json","view_paper":"https://pith.science/paper/DNC3O576","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.00217&json=true","fetch_graph":"https://pith.science/api/pith-number/DNC3O576MEYEGV2JA7LNVO2TDW/graph.json","fetch_events":"https://pith.science/api/pith-number/DNC3O576MEYEGV2JA7LNVO2TDW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW/action/storage_attestation","attest_author":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW/action/author_attestation","sign_citation":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW/action/citation_signature","submit_replication":"https://pith.science/pith/DNC3O576MEYEGV2JA7LNVO2TDW/action/replication_record"}},"created_at":"2026-07-05T10:08:30.666236+00:00","updated_at":"2026-07-05T10:08:30.666236+00:00"}