{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:VGAUTTP2WKC7BR7UR66NX5NNS4","short_pith_number":"pith:VGAUTTP2","schema_version":"1.0","canonical_sha256":"a98149cdfab285f0c7f48fbcdbf5ad9717299a6fd38cfb4c9c535957ab837f39","source":{"kind":"arxiv","id":"2202.02396","version":4},"attestation_state":"computed","paper":{"title":"A Temporal-Difference Approach to Policy Gradient Estimation","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andrew Patterson, A. Rupam Mahmood, Martha White, Samuele Tosatto","submitted_at":"2022-02-04T21:23:33Z","abstract_excerpt":"The policy gradient theorem (Sutton et al., 2000) prescribes the usage of a cumulative discounted state distribution under the target policy to approximate the gradient. Most algorithms based on this theorem, in practice, break this assumption, introducing a distribution shift that can cause the convergence to poor solutions. In this paper, we propose a new approach of reconstructing the policy gradient from the start state without requiring a particular sampling strategy. The policy gradient calculation in this form can be simplified in terms of a gradient critic, which can be recursively est"},"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":"2202.02396","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2022-02-04T21:23:33Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"61423959c0216d14d6c58f4417f0bfc5479f9b5271844e8664e61224258f3e39","abstract_canon_sha256":"4abc957b3621fc93bfd7eb12d3eff1f055e74b0c3f609cc421e2bf9b07ab187e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:38:13.802824Z","signature_b64":"99KsVrRrRvYoMM07NiZweAPggqkgbt0WMnDFoFlAYioSsJWUjSh1/t2nGZR2q5TGlUdodx/Uf3CUsGDE6/fdBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a98149cdfab285f0c7f48fbcdbf5ad9717299a6fd38cfb4c9c535957ab837f39","last_reissued_at":"2026-07-05T04:38:13.802397Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:38:13.802397Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Temporal-Difference Approach to Policy Gradient Estimation","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andrew Patterson, A. Rupam Mahmood, Martha White, Samuele Tosatto","submitted_at":"2022-02-04T21:23:33Z","abstract_excerpt":"The policy gradient theorem (Sutton et al., 2000) prescribes the usage of a cumulative discounted state distribution under the target policy to approximate the gradient. Most algorithms based on this theorem, in practice, break this assumption, introducing a distribution shift that can cause the convergence to poor solutions. In this paper, we propose a new approach of reconstructing the policy gradient from the start state without requiring a particular sampling strategy. The policy gradient calculation in this form can be simplified in terms of a gradient critic, which can be recursively est"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.02396","kind":"arxiv","version":4},"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/2202.02396/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":"2202.02396","created_at":"2026-07-05T04:38:13.802458+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.02396v4","created_at":"2026-07-05T04:38:13.802458+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.02396","created_at":"2026-07-05T04:38:13.802458+00:00"},{"alias_kind":"pith_short_12","alias_value":"VGAUTTP2WKC7","created_at":"2026-07-05T04:38:13.802458+00:00"},{"alias_kind":"pith_short_16","alias_value":"VGAUTTP2WKC7BR7U","created_at":"2026-07-05T04:38:13.802458+00:00"},{"alias_kind":"pith_short_8","alias_value":"VGAUTTP2","created_at":"2026-07-05T04:38:13.802458+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/VGAUTTP2WKC7BR7UR66NX5NNS4","json":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4.json","graph_json":"https://pith.science/api/pith-number/VGAUTTP2WKC7BR7UR66NX5NNS4/graph.json","events_json":"https://pith.science/api/pith-number/VGAUTTP2WKC7BR7UR66NX5NNS4/events.json","paper":"https://pith.science/paper/VGAUTTP2"},"agent_actions":{"view_html":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4","download_json":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4.json","view_paper":"https://pith.science/paper/VGAUTTP2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.02396&json=true","fetch_graph":"https://pith.science/api/pith-number/VGAUTTP2WKC7BR7UR66NX5NNS4/graph.json","fetch_events":"https://pith.science/api/pith-number/VGAUTTP2WKC7BR7UR66NX5NNS4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4/action/storage_attestation","attest_author":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4/action/author_attestation","sign_citation":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4/action/citation_signature","submit_replication":"https://pith.science/pith/VGAUTTP2WKC7BR7UR66NX5NNS4/action/replication_record"}},"created_at":"2026-07-05T04:38:13.802458+00:00","updated_at":"2026-07-05T04:38:13.802458+00:00"}