{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:ZFPF3RVPPFN636UKPT6XH5RMT3","short_pith_number":"pith:ZFPF3RVP","schema_version":"1.0","canonical_sha256":"c95e5dc6af795bedfa8a7cfd73f62c9ef708a69b2951f50e0ef6a47094b17abf","source":{"kind":"arxiv","id":"1805.11592","version":2},"attestation_state":"computed","paper":{"title":"Playing hard exploration games by watching YouTube","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"David Budden, Nando de Freitas, Tobias Pfaff, Tom Le Paine, Yusuf Aytar, Ziyu Wang","submitted_at":"2018-05-29T17:19:36Z","abstract_excerpt":"Deep reinforcement learning methods traditionally struggle with tasks where environment rewards are particularly sparse. One successful method of guiding exploration in these domains is to imitate trajectories provided by a human demonstrator. However, these demonstrations are typically collected under artificial conditions, i.e. with access to the agent's exact environment setup and the demonstrator's action and reward trajectories. Here we propose a two-stage method that overcomes these limitations by relying on noisy, unaligned footage without access to such data. First, we learn to map una"},"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":"1805.11592","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2018-05-29T17:19:36Z","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"title_canon_sha256":"bc4194e2c9cea053d4f994826b7ab6e0d56e209cfcf7b65401114f6251274cef","abstract_canon_sha256":"feba5f7adea23c33ce347dfff8aa10470c263263ba59ebbe88ea543327f41dbc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:59:32.609063Z","signature_b64":"1dUjojpLSid3a36Musb8Kxzrn1GCpyWBx84IyafasqI8id/sgDFaG0iqv1L+WwLCHVuu/skLIUQdBTLb4RIJBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c95e5dc6af795bedfa8a7cfd73f62c9ef708a69b2951f50e0ef6a47094b17abf","last_reissued_at":"2026-05-17T23:59:32.608446Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:59:32.608446Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Playing hard exploration games by watching YouTube","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"David Budden, Nando de Freitas, Tobias Pfaff, Tom Le Paine, Yusuf Aytar, Ziyu Wang","submitted_at":"2018-05-29T17:19:36Z","abstract_excerpt":"Deep reinforcement learning methods traditionally struggle with tasks where environment rewards are particularly sparse. One successful method of guiding exploration in these domains is to imitate trajectories provided by a human demonstrator. However, these demonstrations are typically collected under artificial conditions, i.e. with access to the agent's exact environment setup and the demonstrator's action and reward trajectories. Here we propose a two-stage method that overcomes these limitations by relying on noisy, unaligned footage without access to such data. First, we learn to map una"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1805.11592","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":"1805.11592","created_at":"2026-05-17T23:59:32.608535+00:00"},{"alias_kind":"arxiv_version","alias_value":"1805.11592v2","created_at":"2026-05-17T23:59:32.608535+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1805.11592","created_at":"2026-05-17T23:59:32.608535+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZFPF3RVPPFN6","created_at":"2026-05-18T12:33:07.085635+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZFPF3RVPPFN636UK","created_at":"2026-05-18T12:33:07.085635+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZFPF3RVP","created_at":"2026-05-18T12:33:07.085635+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"1906.11951","citing_title":"Supervise Thyself: Examining Self-Supervised Representations in Interactive Environments","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"1907.02874","citing_title":"Attentive Multi-Task Deep Reinforcement Learning","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2309.16797","citing_title":"Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution","ref_index":35,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3","json":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3.json","graph_json":"https://pith.science/api/pith-number/ZFPF3RVPPFN636UKPT6XH5RMT3/graph.json","events_json":"https://pith.science/api/pith-number/ZFPF3RVPPFN636UKPT6XH5RMT3/events.json","paper":"https://pith.science/paper/ZFPF3RVP"},"agent_actions":{"view_html":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3","download_json":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3.json","view_paper":"https://pith.science/paper/ZFPF3RVP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1805.11592&json=true","fetch_graph":"https://pith.science/api/pith-number/ZFPF3RVPPFN636UKPT6XH5RMT3/graph.json","fetch_events":"https://pith.science/api/pith-number/ZFPF3RVPPFN636UKPT6XH5RMT3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3/action/storage_attestation","attest_author":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3/action/author_attestation","sign_citation":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3/action/citation_signature","submit_replication":"https://pith.science/pith/ZFPF3RVPPFN636UKPT6XH5RMT3/action/replication_record"}},"created_at":"2026-05-17T23:59:32.608535+00:00","updated_at":"2026-05-17T23:59:32.608535+00:00"}