{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:RQJPYBGIWXN6M5HXMVRFWEZO23","short_pith_number":"pith:RQJPYBGI","schema_version":"1.0","canonical_sha256":"8c12fc04c8b5dbe674f765625b132ed6e4d967e7d5de9f5ee0584257b88f397e","source":{"kind":"arxiv","id":"2206.00730","version":3},"attestation_state":"computed","paper":{"title":"The Phenomenon of Policy Churn","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andr\\'e Barreto, Georg Ostrovski, John Quan, Tom Schaul","submitted_at":"2022-06-01T19:44:18Z","abstract_excerpt":"We identify and study the phenomenon of policy churn, that is, the rapid change of the greedy policy in value-based reinforcement learning. Policy churn operates at a surprisingly rapid pace, changing the greedy action in a large fraction of states within a handful of learning updates (in a typical deep RL set-up such as DQN on Atari). We characterise the phenomenon empirically, verifying that it is not limited to specific algorithm or environment properties. A number of ablations help whittle down the plausible explanations on why churn occurs to just a handful, all related to deep learning. "},"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":"2206.00730","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-06-01T19:44:18Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"24eefdd0f3b406c6b883534c984883467adcd05788f690fe572f89d48ea729eb","abstract_canon_sha256":"f4552b774d8cd09a71bf19bc2962b320074a18f5824685af8227422c5331b20b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:08:52.301368Z","signature_b64":"3TACg1s8f6MkiJiiEl+3ZTc8Wgec8DTUT4P3SFkgz+r0i2E9XuxswPDC3IORMFom9ogo8iPlo0xy5oSFiAufBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8c12fc04c8b5dbe674f765625b132ed6e4d967e7d5de9f5ee0584257b88f397e","last_reissued_at":"2026-07-05T05:08:52.300876Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:08:52.300876Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Phenomenon of Policy Churn","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andr\\'e Barreto, Georg Ostrovski, John Quan, Tom Schaul","submitted_at":"2022-06-01T19:44:18Z","abstract_excerpt":"We identify and study the phenomenon of policy churn, that is, the rapid change of the greedy policy in value-based reinforcement learning. Policy churn operates at a surprisingly rapid pace, changing the greedy action in a large fraction of states within a handful of learning updates (in a typical deep RL set-up such as DQN on Atari). We characterise the phenomenon empirically, verifying that it is not limited to specific algorithm or environment properties. A number of ablations help whittle down the plausible explanations on why churn occurs to just a handful, all related to deep learning. "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2206.00730","kind":"arxiv","version":3},"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/2206.00730/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":"2206.00730","created_at":"2026-07-05T05:08:52.300938+00:00"},{"alias_kind":"arxiv_version","alias_value":"2206.00730v3","created_at":"2026-07-05T05:08:52.300938+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2206.00730","created_at":"2026-07-05T05:08:52.300938+00:00"},{"alias_kind":"pith_short_12","alias_value":"RQJPYBGIWXN6","created_at":"2026-07-05T05:08:52.300938+00:00"},{"alias_kind":"pith_short_16","alias_value":"RQJPYBGIWXN6M5HX","created_at":"2026-07-05T05:08:52.300938+00:00"},{"alias_kind":"pith_short_8","alias_value":"RQJPYBGI","created_at":"2026-07-05T05:08:52.300938+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.00592","citing_title":"Mitigating Plasticity Loss in Continual Reinforcement Learning by Reducing Churn","ref_index":2019,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23","json":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23.json","graph_json":"https://pith.science/api/pith-number/RQJPYBGIWXN6M5HXMVRFWEZO23/graph.json","events_json":"https://pith.science/api/pith-number/RQJPYBGIWXN6M5HXMVRFWEZO23/events.json","paper":"https://pith.science/paper/RQJPYBGI"},"agent_actions":{"view_html":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23","download_json":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23.json","view_paper":"https://pith.science/paper/RQJPYBGI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2206.00730&json=true","fetch_graph":"https://pith.science/api/pith-number/RQJPYBGIWXN6M5HXMVRFWEZO23/graph.json","fetch_events":"https://pith.science/api/pith-number/RQJPYBGIWXN6M5HXMVRFWEZO23/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23/action/storage_attestation","attest_author":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23/action/author_attestation","sign_citation":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23/action/citation_signature","submit_replication":"https://pith.science/pith/RQJPYBGIWXN6M5HXMVRFWEZO23/action/replication_record"}},"created_at":"2026-07-05T05:08:52.300938+00:00","updated_at":"2026-07-05T05:08:52.300938+00:00"}