{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:YPSFYXZLSUMOEJ3K4L7GHW3US5","short_pith_number":"pith:YPSFYXZL","schema_version":"1.0","canonical_sha256":"c3e45c5f2b9518e2276ae2fe63db7497708be14d80e952dfe51bacfee3de596f","source":{"kind":"arxiv","id":"2303.07507","version":1},"attestation_state":"computed","paper":{"title":"Loss of Plasticity in Continual Deep Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adam White, Joseph Modayil, Marlos C. Machado, Rosie Zhao, Zaheer Abbas","submitted_at":"2023-03-13T22:37:15Z","abstract_excerpt":"The ability to learn continually is essential in a complex and changing world. In this paper, we characterize the behavior of canonical value-based deep reinforcement learning (RL) approaches under varying degrees of non-stationarity. In particular, we demonstrate that deep RL agents lose their ability to learn good policies when they cycle through a sequence of Atari 2600 games. This phenomenon is alluded to in prior work under various guises -- e.g., loss of plasticity, implicit under-parameterization, primacy bias, and capacity loss. We investigate this phenomenon closely at scale and analy"},"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":"2303.07507","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-03-13T22:37:15Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1222edc23e40a7b4119d2ac34be76d5a23233934cd7a1c64b3575d80f3ac7e95","abstract_canon_sha256":"38c4ac93174fdbc7726d66c1b88c3c0c0e3b60db1e25e9203a3340ff61636b2e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:51:06.819402Z","signature_b64":"P9c4XUaGpaR2gurAKsyUFRU7mR7hDpTBHn9vmKkaG1mVFDK/gGMqvNATZkrDLpPoQPVefP8wU9feJpqMjS0aBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c3e45c5f2b9518e2276ae2fe63db7497708be14d80e952dfe51bacfee3de596f","last_reissued_at":"2026-07-05T05:51:06.818881Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:51:06.818881Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Loss of Plasticity in Continual Deep Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adam White, Joseph Modayil, Marlos C. Machado, Rosie Zhao, Zaheer Abbas","submitted_at":"2023-03-13T22:37:15Z","abstract_excerpt":"The ability to learn continually is essential in a complex and changing world. In this paper, we characterize the behavior of canonical value-based deep reinforcement learning (RL) approaches under varying degrees of non-stationarity. In particular, we demonstrate that deep RL agents lose their ability to learn good policies when they cycle through a sequence of Atari 2600 games. This phenomenon is alluded to in prior work under various guises -- e.g., loss of plasticity, implicit under-parameterization, primacy bias, and capacity loss. We investigate this phenomenon closely at scale and analy"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2303.07507","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/2303.07507/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":"2303.07507","created_at":"2026-07-05T05:51:06.818945+00:00"},{"alias_kind":"arxiv_version","alias_value":"2303.07507v1","created_at":"2026-07-05T05:51:06.818945+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.07507","created_at":"2026-07-05T05:51:06.818945+00:00"},{"alias_kind":"pith_short_12","alias_value":"YPSFYXZLSUMO","created_at":"2026-07-05T05:51:06.818945+00:00"},{"alias_kind":"pith_short_16","alias_value":"YPSFYXZLSUMOEJ3K","created_at":"2026-07-05T05:51:06.818945+00:00"},{"alias_kind":"pith_short_8","alias_value":"YPSFYXZL","created_at":"2026-07-05T05:51:06.818945+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.09028","citing_title":"Plasticity-Enhanced Multi-Agent Mixture of Experts for Dynamic Objective Adaptation in UAVs-Assisted Emergency Communication Networks","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5","json":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5.json","graph_json":"https://pith.science/api/pith-number/YPSFYXZLSUMOEJ3K4L7GHW3US5/graph.json","events_json":"https://pith.science/api/pith-number/YPSFYXZLSUMOEJ3K4L7GHW3US5/events.json","paper":"https://pith.science/paper/YPSFYXZL"},"agent_actions":{"view_html":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5","download_json":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5.json","view_paper":"https://pith.science/paper/YPSFYXZL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2303.07507&json=true","fetch_graph":"https://pith.science/api/pith-number/YPSFYXZLSUMOEJ3K4L7GHW3US5/graph.json","fetch_events":"https://pith.science/api/pith-number/YPSFYXZLSUMOEJ3K4L7GHW3US5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5/action/storage_attestation","attest_author":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5/action/author_attestation","sign_citation":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5/action/citation_signature","submit_replication":"https://pith.science/pith/YPSFYXZLSUMOEJ3K4L7GHW3US5/action/replication_record"}},"created_at":"2026-07-05T05:51:06.818945+00:00","updated_at":"2026-07-05T05:51:06.818945+00:00"}