{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:DLBMWF53RK5QVN2YSZ36HECWA4","short_pith_number":"pith:DLBMWF53","schema_version":"1.0","canonical_sha256":"1ac2cb17bb8abb0ab7589677e39056070c0c1f90f5b58ef244e595256b455725","source":{"kind":"arxiv","id":"2405.09999","version":2},"attestation_state":"computed","paper":{"title":"Reward Centering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Abhishek Naik, Manan Tomar, Richard S. Sutton, Yi Wan","submitted_at":"2024-05-16T11:33:49Z","abstract_excerpt":"We show that discounted methods for solving continuing reinforcement learning problems can perform significantly better if they center their rewards by subtracting out the rewards' empirical average. The improvement is substantial at commonly used discount factors and increases further as the discount factor approaches one. In addition, we show that if a problem's rewards are shifted by a constant, then standard methods perform much worse, whereas methods with reward centering are unaffected. Estimating the average reward is straightforward in the on-policy setting; we propose a slightly more "},"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":"2405.09999","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-16T11:33:49Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8bb3409356ffc1706e973be7e3c9aee0572c650093b9df20efcc6887356763f1","abstract_canon_sha256":"e2ae4638c34d622df660cdc352a2813de791132e8cc787cdbedebeb4e5b901ac"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:28:30.877397Z","signature_b64":"9naRqBDUuq37SRL3CWjLDjCq8LeE9Jsx5TMQHTLEpvayldDUOaxEp4W6I7VtdnK99WFkMm82Fht/2PHfvoXUAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1ac2cb17bb8abb0ab7589677e39056070c0c1f90f5b58ef244e595256b455725","last_reissued_at":"2026-07-05T09:28:30.876914Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:28:30.876914Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reward Centering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Abhishek Naik, Manan Tomar, Richard S. Sutton, Yi Wan","submitted_at":"2024-05-16T11:33:49Z","abstract_excerpt":"We show that discounted methods for solving continuing reinforcement learning problems can perform significantly better if they center their rewards by subtracting out the rewards' empirical average. The improvement is substantial at commonly used discount factors and increases further as the discount factor approaches one. In addition, we show that if a problem's rewards are shifted by a constant, then standard methods perform much worse, whereas methods with reward centering are unaffected. Estimating the average reward is straightforward in the on-policy setting; we propose a slightly more "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.09999","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/2405.09999/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":"2405.09999","created_at":"2026-07-05T09:28:30.876965+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.09999v2","created_at":"2026-07-05T09:28:30.876965+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.09999","created_at":"2026-07-05T09:28:30.876965+00:00"},{"alias_kind":"pith_short_12","alias_value":"DLBMWF53RK5Q","created_at":"2026-07-05T09:28:30.876965+00:00"},{"alias_kind":"pith_short_16","alias_value":"DLBMWF53RK5QVN2Y","created_at":"2026-07-05T09:28:30.876965+00:00"},{"alias_kind":"pith_short_8","alias_value":"DLBMWF53","created_at":"2026-07-05T09:28:30.876965+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.02075","citing_title":"Graph Transformers and Stabilized Reinforcement Learning for Large-Scale Dynamic Routing Modulation and Spectrum Allocation in Elastic Optical Networks","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04539","citing_title":"FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04539","citing_title":"FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02075","citing_title":"Graph Transformers and Stabilized Reinforcement Learning for Large-Scale Dynamic Routing Modulation and Spectrum Allocation in Elastic Optical Networks","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4","json":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4.json","graph_json":"https://pith.science/api/pith-number/DLBMWF53RK5QVN2YSZ36HECWA4/graph.json","events_json":"https://pith.science/api/pith-number/DLBMWF53RK5QVN2YSZ36HECWA4/events.json","paper":"https://pith.science/paper/DLBMWF53"},"agent_actions":{"view_html":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4","download_json":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4.json","view_paper":"https://pith.science/paper/DLBMWF53","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.09999&json=true","fetch_graph":"https://pith.science/api/pith-number/DLBMWF53RK5QVN2YSZ36HECWA4/graph.json","fetch_events":"https://pith.science/api/pith-number/DLBMWF53RK5QVN2YSZ36HECWA4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4/action/storage_attestation","attest_author":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4/action/author_attestation","sign_citation":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4/action/citation_signature","submit_replication":"https://pith.science/pith/DLBMWF53RK5QVN2YSZ36HECWA4/action/replication_record"}},"created_at":"2026-07-05T09:28:30.876965+00:00","updated_at":"2026-07-05T09:28:30.876965+00:00"}