{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:4DEANPA6VVJZ2J6AAL4YCPHTGA","short_pith_number":"pith:4DEANPA6","schema_version":"1.0","canonical_sha256":"e0c806bc1ead539d27c002f9813cf3301d16106d9863903a02f0f73028b96957","source":{"kind":"arxiv","id":"1703.05449","version":2},"attestation_state":"computed","paper":{"title":"Minimax Regret Bounds for Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Ian Osband, Mohammad Gheshlaghi Azar, R\\'emi Munos","submitted_at":"2017-03-16T01:31:33Z","abstract_excerpt":"We consider the problem of provably optimal exploration in reinforcement learning for finite horizon MDPs. We show that an optimistic modification to value iteration achieves a regret bound of $\\tilde{O}( \\sqrt{HSAT} + H^2S^2A+H\\sqrt{T})$ where $H$ is the time horizon, $S$ the number of states, $A$ the number of actions and $T$ the number of time-steps. This result improves over the best previous known bound $\\tilde{O}(HS \\sqrt{AT})$ achieved by the UCRL2 algorithm of Jaksch et al., 2010. The key significance of our new results is that when $T\\geq H^3S^3A$ and $SA\\geq H$, it leads to a regret "},"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":"1703.05449","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2017-03-16T01:31:33Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"58128e741541cf1c8bf6d5b8fe61ff473c2ae9a2ed260604d7355da6f3ab6cad","abstract_canon_sha256":"76d4ea69425fca02e9507e905c1b0dcb856733c785ac241d44f210170c3715b0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:41:07.231301Z","signature_b64":"3zjfhApMvV+c2HgvAFkJKSmAjOpkOhMlmlMDUQoImyI97cts3JrIIGRVylI2OjCnn+QOrCCY2ieYef7/neuKDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e0c806bc1ead539d27c002f9813cf3301d16106d9863903a02f0f73028b96957","last_reissued_at":"2026-05-18T00:41:07.230577Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:41:07.230577Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Minimax Regret Bounds for Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Ian Osband, Mohammad Gheshlaghi Azar, R\\'emi Munos","submitted_at":"2017-03-16T01:31:33Z","abstract_excerpt":"We consider the problem of provably optimal exploration in reinforcement learning for finite horizon MDPs. We show that an optimistic modification to value iteration achieves a regret bound of $\\tilde{O}( \\sqrt{HSAT} + H^2S^2A+H\\sqrt{T})$ where $H$ is the time horizon, $S$ the number of states, $A$ the number of actions and $T$ the number of time-steps. This result improves over the best previous known bound $\\tilde{O}(HS \\sqrt{AT})$ achieved by the UCRL2 algorithm of Jaksch et al., 2010. The key significance of our new results is that when $T\\geq H^3S^3A$ and $SA\\geq H$, it leads to a regret "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1703.05449","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":""},"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":"1703.05449","created_at":"2026-05-18T00:41:07.230697+00:00"},{"alias_kind":"arxiv_version","alias_value":"1703.05449v2","created_at":"2026-05-18T00:41:07.230697+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1703.05449","created_at":"2026-05-18T00:41:07.230697+00:00"},{"alias_kind":"pith_short_12","alias_value":"4DEANPA6VVJZ","created_at":"2026-05-18T12:30:58.224056+00:00"},{"alias_kind":"pith_short_16","alias_value":"4DEANPA6VVJZ2J6A","created_at":"2026-05-18T12:30:58.224056+00:00"},{"alias_kind":"pith_short_8","alias_value":"4DEANPA6","created_at":"2026-05-18T12:30:58.224056+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"1906.09114","citing_title":"Near-optimal Bayesian Solution For Unknown Discrete Markov Decision Process","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2509.25424","citing_title":"Polychromic Objectives for Reinforcement Learning","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2605.00393","citing_title":"Model-Based Reinforcement Learning with Double Oracle Efficiency in Policy Optimization and Offline Estimation","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA","json":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA.json","graph_json":"https://pith.science/api/pith-number/4DEANPA6VVJZ2J6AAL4YCPHTGA/graph.json","events_json":"https://pith.science/api/pith-number/4DEANPA6VVJZ2J6AAL4YCPHTGA/events.json","paper":"https://pith.science/paper/4DEANPA6"},"agent_actions":{"view_html":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA","download_json":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA.json","view_paper":"https://pith.science/paper/4DEANPA6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1703.05449&json=true","fetch_graph":"https://pith.science/api/pith-number/4DEANPA6VVJZ2J6AAL4YCPHTGA/graph.json","fetch_events":"https://pith.science/api/pith-number/4DEANPA6VVJZ2J6AAL4YCPHTGA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA/action/storage_attestation","attest_author":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA/action/author_attestation","sign_citation":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA/action/citation_signature","submit_replication":"https://pith.science/pith/4DEANPA6VVJZ2J6AAL4YCPHTGA/action/replication_record"}},"created_at":"2026-05-18T00:41:07.230697+00:00","updated_at":"2026-05-18T00:41:07.230697+00:00"}