{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:AKZER62NHSI4CVKDVXQABBZ5HP","short_pith_number":"pith:AKZER62N","schema_version":"1.0","canonical_sha256":"02b248fb4d3c91c15543ade000873d3bc31ce529180ca0c71ab13777f17319ab","source":{"kind":"arxiv","id":"2002.12174","version":1},"attestation_state":"computed","paper":{"title":"Optimistic Exploration even with a Pessimistic Initialisation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Bei Peng, Shimon Whiteson, Tabish Rashid, Wendelin B\\\"ohmer","submitted_at":"2020-02-26T17:15:53Z","abstract_excerpt":"Optimistic initialisation is an effective strategy for efficient exploration in reinforcement learning (RL). In the tabular case, all provably efficient model-free algorithms rely on it. However, model-free deep RL algorithms do not use optimistic initialisation despite taking inspiration from these provably efficient tabular algorithms. In particular, in scenarios with only positive rewards, Q-values are initialised at their lowest possible values due to commonly used network initialisation schemes, a pessimistic initialisation. Merely initialising the network to output optimistic Q-values is"},"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":"2002.12174","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-02-26T17:15:53Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"1a87eb0e1d19b845daf8e6896636e5efa90c5558f0a7b2f6f9c8d2827c69ce9b","abstract_canon_sha256":"262b43479ac50e3acf4a7255c7f10ebf9f22a482e6ed4da7418e6ccb73b23458"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:44:21.176930Z","signature_b64":"wPZECVq3D7CYW8JfF1LS1rYVU4aubbBDCSPRpc4Jl7fxw2vhN4BD5GIKskWj8PFt4iWukuGvdqqGYcFytntYCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"02b248fb4d3c91c15543ade000873d3bc31ce529180ca0c71ab13777f17319ab","last_reissued_at":"2026-07-05T00:44:21.176531Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:44:21.176531Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Optimistic Exploration even with a Pessimistic Initialisation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Bei Peng, Shimon Whiteson, Tabish Rashid, Wendelin B\\\"ohmer","submitted_at":"2020-02-26T17:15:53Z","abstract_excerpt":"Optimistic initialisation is an effective strategy for efficient exploration in reinforcement learning (RL). In the tabular case, all provably efficient model-free algorithms rely on it. However, model-free deep RL algorithms do not use optimistic initialisation despite taking inspiration from these provably efficient tabular algorithms. In particular, in scenarios with only positive rewards, Q-values are initialised at their lowest possible values due to commonly used network initialisation schemes, a pessimistic initialisation. Merely initialising the network to output optimistic Q-values is"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2002.12174","kind":"arxiv","version":1},"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/2002.12174/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":"2002.12174","created_at":"2026-07-05T00:44:21.176596+00:00"},{"alias_kind":"arxiv_version","alias_value":"2002.12174v1","created_at":"2026-07-05T00:44:21.176596+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2002.12174","created_at":"2026-07-05T00:44:21.176596+00:00"},{"alias_kind":"pith_short_12","alias_value":"AKZER62NHSI4","created_at":"2026-07-05T00:44:21.176596+00:00"},{"alias_kind":"pith_short_16","alias_value":"AKZER62NHSI4CVKD","created_at":"2026-07-05T00:44:21.176596+00:00"},{"alias_kind":"pith_short_8","alias_value":"AKZER62N","created_at":"2026-07-05T00:44:21.176596+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.13672","citing_title":"The Courage to Stop: Overcoming Sunk Cost Fallacy in Deep Reinforcement Learning","ref_index":48,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP","json":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP.json","graph_json":"https://pith.science/api/pith-number/AKZER62NHSI4CVKDVXQABBZ5HP/graph.json","events_json":"https://pith.science/api/pith-number/AKZER62NHSI4CVKDVXQABBZ5HP/events.json","paper":"https://pith.science/paper/AKZER62N"},"agent_actions":{"view_html":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP","download_json":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP.json","view_paper":"https://pith.science/paper/AKZER62N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2002.12174&json=true","fetch_graph":"https://pith.science/api/pith-number/AKZER62NHSI4CVKDVXQABBZ5HP/graph.json","fetch_events":"https://pith.science/api/pith-number/AKZER62NHSI4CVKDVXQABBZ5HP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP/action/storage_attestation","attest_author":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP/action/author_attestation","sign_citation":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP/action/citation_signature","submit_replication":"https://pith.science/pith/AKZER62NHSI4CVKDVXQABBZ5HP/action/replication_record"}},"created_at":"2026-07-05T00:44:21.176596+00:00","updated_at":"2026-07-05T00:44:21.176596+00:00"}