{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:FT24MCKQLEUVKNRSIM55E5K6DA","short_pith_number":"pith:FT24MCKQ","schema_version":"1.0","canonical_sha256":"2cf5c609505929553632433bd2755e18129e25ed693c4f272c813b89e4f52792","source":{"kind":"arxiv","id":"1904.04025","version":5},"attestation_state":"computed","paper":{"title":"Only Relevant Information Matters: Filtering Out Noisy Samples to Boost RL","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Philippe Preux, Yannis Flet-Berliac","submitted_at":"2019-04-08T12:53:12Z","abstract_excerpt":"In reinforcement learning, policy gradient algorithms optimize the policy directly and rely on sampling efficiently an environment. Nevertheless, while most sampling procedures are based on direct policy sampling, self-performance measures could be used to improve such sampling prior to each policy update. Following this line of thought, we introduce SAUNA, a method where non-informative transitions are rejected from the gradient update. The level of information is estimated according to the fraction of variance explained by the value function: a measure of the discrepancy between V and the em"},"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":"1904.04025","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-04-08T12:53:12Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"1d251fd52716f248a0699dafe4507a8e1785a357edcff4859d3d9e16bd630be3","abstract_canon_sha256":"c03a0632f1b89fb63b5e35eca14c46d7a0b271340bc9a51530fca3c2726fb199"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:53:06.353339Z","signature_b64":"UfWJ2grp/2zNncdv57KlIgPiM71d+cTSbOTu2dCK+f0/uJWwMdfLOrtLP9ckB/bSTYp+kG8q+wljGHbT33EbBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2cf5c609505929553632433bd2755e18129e25ed693c4f272c813b89e4f52792","last_reissued_at":"2026-07-05T01:53:06.352796Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:53:06.352796Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Only Relevant Information Matters: Filtering Out Noisy Samples to Boost RL","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Philippe Preux, Yannis Flet-Berliac","submitted_at":"2019-04-08T12:53:12Z","abstract_excerpt":"In reinforcement learning, policy gradient algorithms optimize the policy directly and rely on sampling efficiently an environment. Nevertheless, while most sampling procedures are based on direct policy sampling, self-performance measures could be used to improve such sampling prior to each policy update. Following this line of thought, we introduce SAUNA, a method where non-informative transitions are rejected from the gradient update. The level of information is estimated according to the fraction of variance explained by the value function: a measure of the discrepancy between V and the em"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1904.04025","kind":"arxiv","version":5},"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/1904.04025/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":"1904.04025","created_at":"2026-07-05T01:53:06.352855+00:00"},{"alias_kind":"arxiv_version","alias_value":"1904.04025v5","created_at":"2026-07-05T01:53:06.352855+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1904.04025","created_at":"2026-07-05T01:53:06.352855+00:00"},{"alias_kind":"pith_short_12","alias_value":"FT24MCKQLEUV","created_at":"2026-07-05T01:53:06.352855+00:00"},{"alias_kind":"pith_short_16","alias_value":"FT24MCKQLEUVKNRS","created_at":"2026-07-05T01:53:06.352855+00:00"},{"alias_kind":"pith_short_8","alias_value":"FT24MCKQ","created_at":"2026-07-05T01:53:06.352855+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.19485","citing_title":"EVPO: Explained Variance Policy Optimization for Adaptive Critic Utilization in LLM Post-Training","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA","json":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA.json","graph_json":"https://pith.science/api/pith-number/FT24MCKQLEUVKNRSIM55E5K6DA/graph.json","events_json":"https://pith.science/api/pith-number/FT24MCKQLEUVKNRSIM55E5K6DA/events.json","paper":"https://pith.science/paper/FT24MCKQ"},"agent_actions":{"view_html":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA","download_json":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA.json","view_paper":"https://pith.science/paper/FT24MCKQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1904.04025&json=true","fetch_graph":"https://pith.science/api/pith-number/FT24MCKQLEUVKNRSIM55E5K6DA/graph.json","fetch_events":"https://pith.science/api/pith-number/FT24MCKQLEUVKNRSIM55E5K6DA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA/action/storage_attestation","attest_author":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA/action/author_attestation","sign_citation":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA/action/citation_signature","submit_replication":"https://pith.science/pith/FT24MCKQLEUVKNRSIM55E5K6DA/action/replication_record"}},"created_at":"2026-07-05T01:53:06.352855+00:00","updated_at":"2026-07-05T01:53:06.352855+00:00"}