{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:YBXREELNEWFHEYLG7FUGULSLGO","short_pith_number":"pith:YBXREELN","schema_version":"1.0","canonical_sha256":"c06f12116d258a726166f9686a2e4b33b04789c32e0aa7e568020caaf3b78a65","source":{"kind":"arxiv","id":"1912.02975","version":2},"attestation_state":"computed","paper":{"title":"Observational Overfitting in Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Behnam Neyshabur, Stephen Tu, Xingyou Song, Yiding Jiang, Yilun Du","submitted_at":"2019-12-06T04:52:16Z","abstract_excerpt":"A major component of overfitting in model-free reinforcement learning (RL) involves the case where the agent may mistakenly correlate reward with certain spurious features from the observations generated by the Markov Decision Process (MDP). We provide a general framework for analyzing this scenario, which we use to design multiple synthetic benchmarks from only modifying the observation space of an MDP. When an agent overfits to different observation spaces even if the underlying MDP dynamics is fixed, we term this observational overfitting. Our experiments expose intriguing properties especi"},"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":"1912.02975","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-12-06T04:52:16Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"e182c4382f8b62e9be09c7937c446c11d91fdbc6a716b3b2362da77e4594dc29","abstract_canon_sha256":"e8cc0523c905a43f9dcdef242ce2b2167bd5c90dc3795bbacda1e5012ac1f36f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:28:57.472048Z","signature_b64":"zDsAiQDwPSzcMkR97QdPr2gOtkMWkHkmeXmvhJEoVUpxNdteRFAWgpSdPnUpJ0OCpGUw49uW4xdCN1i3+GWiDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c06f12116d258a726166f9686a2e4b33b04789c32e0aa7e568020caaf3b78a65","last_reissued_at":"2026-07-05T00:28:57.471637Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:28:57.471637Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Observational Overfitting in Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Behnam Neyshabur, Stephen Tu, Xingyou Song, Yiding Jiang, Yilun Du","submitted_at":"2019-12-06T04:52:16Z","abstract_excerpt":"A major component of overfitting in model-free reinforcement learning (RL) involves the case where the agent may mistakenly correlate reward with certain spurious features from the observations generated by the Markov Decision Process (MDP). We provide a general framework for analyzing this scenario, which we use to design multiple synthetic benchmarks from only modifying the observation space of an MDP. When an agent overfits to different observation spaces even if the underlying MDP dynamics is fixed, we term this observational overfitting. Our experiments expose intriguing properties especi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1912.02975","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/1912.02975/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":"1912.02975","created_at":"2026-07-05T00:28:57.471696+00:00"},{"alias_kind":"arxiv_version","alias_value":"1912.02975v2","created_at":"2026-07-05T00:28:57.471696+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1912.02975","created_at":"2026-07-05T00:28:57.471696+00:00"},{"alias_kind":"pith_short_12","alias_value":"YBXREELNEWFH","created_at":"2026-07-05T00:28:57.471696+00:00"},{"alias_kind":"pith_short_16","alias_value":"YBXREELNEWFHEYLG","created_at":"2026-07-05T00:28:57.471696+00:00"},{"alias_kind":"pith_short_8","alias_value":"YBXREELN","created_at":"2026-07-05T00:28:57.471696+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2210.10760","citing_title":"Scaling Laws for Reward Model Overoptimization","ref_index":28,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO","json":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO.json","graph_json":"https://pith.science/api/pith-number/YBXREELNEWFHEYLG7FUGULSLGO/graph.json","events_json":"https://pith.science/api/pith-number/YBXREELNEWFHEYLG7FUGULSLGO/events.json","paper":"https://pith.science/paper/YBXREELN"},"agent_actions":{"view_html":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO","download_json":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO.json","view_paper":"https://pith.science/paper/YBXREELN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1912.02975&json=true","fetch_graph":"https://pith.science/api/pith-number/YBXREELNEWFHEYLG7FUGULSLGO/graph.json","fetch_events":"https://pith.science/api/pith-number/YBXREELNEWFHEYLG7FUGULSLGO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO/action/storage_attestation","attest_author":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO/action/author_attestation","sign_citation":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO/action/citation_signature","submit_replication":"https://pith.science/pith/YBXREELNEWFHEYLG7FUGULSLGO/action/replication_record"}},"created_at":"2026-07-05T00:28:57.471696+00:00","updated_at":"2026-07-05T00:28:57.471696+00:00"}