{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:PN42TRUCPLLYHUK6KTUPHKUU4C","short_pith_number":"pith:PN42TRUC","schema_version":"1.0","canonical_sha256":"7b79a9c6827ad783d15e54e8f3aa94e0a06f8272d14d4f71d7eed784417503a2","source":{"kind":"arxiv","id":"2111.14552","version":2},"attestation_state":"computed","paper":{"title":"Robust On-Policy Sampling for Data-Efficient Policy Evaluation in Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Duohan Zhang, Josiah P. Hanna, Lukas Sch\\\"afer, Rujie Zhong, Stefano V. Albrecht","submitted_at":"2021-11-29T14:30:26Z","abstract_excerpt":"Reinforcement learning (RL) algorithms are often categorized as either on-policy or off-policy depending on whether they use data from a target policy of interest or from a different behavior policy. In this paper, we study a subtle distinction between on-policy data and on-policy sampling in the context of the RL sub-problem of policy evaluation. We observe that on-policy sampling may fail to match the expected distribution of on-policy data after observing only a finite number of trajectories and this failure hinders data-efficient policy evaluation. Towards improved data-efficiency, we show"},"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":"2111.14552","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-11-29T14:30:26Z","cross_cats_sorted":[],"title_canon_sha256":"0d0c0a44304c8f91c16a60bd7a5805ff65b62a0424529cc1661e2428610c3d8e","abstract_canon_sha256":"607abda45fc84398dbbbb1164caef13976a1bf213cafb9d159be055311f16e56"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:05:01.879116Z","signature_b64":"vG2B6fzkzfaDoKARNGHJrNZcA7sYRbu7VTqOWv/JYC+OGYnwuWtxv4l5n/NCwAZin3MkuRt96Sy/nHzBmQZ5Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7b79a9c6827ad783d15e54e8f3aa94e0a06f8272d14d4f71d7eed784417503a2","last_reissued_at":"2026-07-05T05:05:01.878612Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:05:01.878612Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robust On-Policy Sampling for Data-Efficient Policy Evaluation in Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Duohan Zhang, Josiah P. Hanna, Lukas Sch\\\"afer, Rujie Zhong, Stefano V. Albrecht","submitted_at":"2021-11-29T14:30:26Z","abstract_excerpt":"Reinforcement learning (RL) algorithms are often categorized as either on-policy or off-policy depending on whether they use data from a target policy of interest or from a different behavior policy. In this paper, we study a subtle distinction between on-policy data and on-policy sampling in the context of the RL sub-problem of policy evaluation. We observe that on-policy sampling may fail to match the expected distribution of on-policy data after observing only a finite number of trajectories and this failure hinders data-efficient policy evaluation. Towards improved data-efficiency, we show"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.14552","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/2111.14552/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":"2111.14552","created_at":"2026-07-05T05:05:01.878676+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.14552v2","created_at":"2026-07-05T05:05:01.878676+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.14552","created_at":"2026-07-05T05:05:01.878676+00:00"},{"alias_kind":"pith_short_12","alias_value":"PN42TRUCPLLY","created_at":"2026-07-05T05:05:01.878676+00:00"},{"alias_kind":"pith_short_16","alias_value":"PN42TRUCPLLYHUK6","created_at":"2026-07-05T05:05:01.878676+00:00"},{"alias_kind":"pith_short_8","alias_value":"PN42TRUC","created_at":"2026-07-05T05:05:01.878676+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/PN42TRUCPLLYHUK6KTUPHKUU4C","json":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C.json","graph_json":"https://pith.science/api/pith-number/PN42TRUCPLLYHUK6KTUPHKUU4C/graph.json","events_json":"https://pith.science/api/pith-number/PN42TRUCPLLYHUK6KTUPHKUU4C/events.json","paper":"https://pith.science/paper/PN42TRUC"},"agent_actions":{"view_html":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C","download_json":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C.json","view_paper":"https://pith.science/paper/PN42TRUC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.14552&json=true","fetch_graph":"https://pith.science/api/pith-number/PN42TRUCPLLYHUK6KTUPHKUU4C/graph.json","fetch_events":"https://pith.science/api/pith-number/PN42TRUCPLLYHUK6KTUPHKUU4C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C/action/storage_attestation","attest_author":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C/action/author_attestation","sign_citation":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C/action/citation_signature","submit_replication":"https://pith.science/pith/PN42TRUCPLLYHUK6KTUPHKUU4C/action/replication_record"}},"created_at":"2026-07-05T05:05:01.878676+00:00","updated_at":"2026-07-05T05:05:01.878676+00:00"}