{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:K74ISTEGZ2GM6YZFS5MPG22QJT","short_pith_number":"pith:K74ISTEG","schema_version":"1.0","canonical_sha256":"57f8894c86ce8ccf63259758f36b504cd44cd9d15110163a133a18f2135cc1c2","source":{"kind":"arxiv","id":"2205.11027","version":3},"attestation_state":"computed","paper":{"title":"When Data Geometry Meets Deep Function: Generalizing Offline Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Haoran Xu, Jianxiong Li, Jingjing Liu, Xiangyu Zhu, Xianyuan Zhan, Ya-Qin Zhang","submitted_at":"2022-05-23T04:01:11Z","abstract_excerpt":"In offline reinforcement learning (RL), one detrimental issue to policy learning is the error accumulation of deep Q function in out-of-distribution (OOD) areas. Unfortunately, existing offline RL methods are often over-conservative, inevitably hurting generalization performance outside data distribution. In our study, one interesting observation is that deep Q functions approximate well inside the convex hull of training data. Inspired by this, we propose a new method, DOGE (Distance-sensitive Offline RL with better GEneralization). DOGE marries dataset geometry with deep function approximato"},"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":"2205.11027","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-05-23T04:01:11Z","cross_cats_sorted":["cs.AI","cs.RO"],"title_canon_sha256":"324c00503e3fe2f6c4312e213e77b6ae66b54f569b654d2f69a0e60ae1e1a092","abstract_canon_sha256":"294738af4fd2f23801ed7b41343f5ab881833dff6d5dbebe90f5c8ded80b3b9e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:46:51.521769Z","signature_b64":"LYpveNaGz/cA0zUNgFGHPBQnvg3mjocvSmz5FbymlQ6iZvon+af+hu1XwO1cqOztavzmm9HGCzfm43TXxNwmBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"57f8894c86ce8ccf63259758f36b504cd44cd9d15110163a133a18f2135cc1c2","last_reissued_at":"2026-07-05T05:46:51.521348Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:46:51.521348Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When Data Geometry Meets Deep Function: Generalizing Offline Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Haoran Xu, Jianxiong Li, Jingjing Liu, Xiangyu Zhu, Xianyuan Zhan, Ya-Qin Zhang","submitted_at":"2022-05-23T04:01:11Z","abstract_excerpt":"In offline reinforcement learning (RL), one detrimental issue to policy learning is the error accumulation of deep Q function in out-of-distribution (OOD) areas. Unfortunately, existing offline RL methods are often over-conservative, inevitably hurting generalization performance outside data distribution. In our study, one interesting observation is that deep Q functions approximate well inside the convex hull of training data. Inspired by this, we propose a new method, DOGE (Distance-sensitive Offline RL with better GEneralization). DOGE marries dataset geometry with deep function approximato"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.11027","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/2205.11027/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":"2205.11027","created_at":"2026-07-05T05:46:51.521409+00:00"},{"alias_kind":"arxiv_version","alias_value":"2205.11027v3","created_at":"2026-07-05T05:46:51.521409+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.11027","created_at":"2026-07-05T05:46:51.521409+00:00"},{"alias_kind":"pith_short_12","alias_value":"K74ISTEGZ2GM","created_at":"2026-07-05T05:46:51.521409+00:00"},{"alias_kind":"pith_short_16","alias_value":"K74ISTEGZ2GM6YZF","created_at":"2026-07-05T05:46:51.521409+00:00"},{"alias_kind":"pith_short_8","alias_value":"K74ISTEG","created_at":"2026-07-05T05:46:51.521409+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.08417","citing_title":"Offline RL with Smooth OOD Generalization in Convex Hull and its Neighborhood","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT","json":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT.json","graph_json":"https://pith.science/api/pith-number/K74ISTEGZ2GM6YZFS5MPG22QJT/graph.json","events_json":"https://pith.science/api/pith-number/K74ISTEGZ2GM6YZFS5MPG22QJT/events.json","paper":"https://pith.science/paper/K74ISTEG"},"agent_actions":{"view_html":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT","download_json":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT.json","view_paper":"https://pith.science/paper/K74ISTEG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2205.11027&json=true","fetch_graph":"https://pith.science/api/pith-number/K74ISTEGZ2GM6YZFS5MPG22QJT/graph.json","fetch_events":"https://pith.science/api/pith-number/K74ISTEGZ2GM6YZFS5MPG22QJT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT/action/storage_attestation","attest_author":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT/action/author_attestation","sign_citation":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT/action/citation_signature","submit_replication":"https://pith.science/pith/K74ISTEGZ2GM6YZFS5MPG22QJT/action/replication_record"}},"created_at":"2026-07-05T05:46:51.521409+00:00","updated_at":"2026-07-05T05:46:51.521409+00:00"}