{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AR4TMWBE374PW2ZO3CMSVEIAN5","short_pith_number":"pith:AR4TMWBE","schema_version":"1.0","canonical_sha256":"0479365824dff8fb6b2ed8992a91006f79e5a3c9b946a0616855ee25b842ffdb","source":{"kind":"arxiv","id":"2606.00151","version":1},"attestation_state":"computed","paper":{"title":"Emergence of Exploration in Policy Gradient Reinforcement Learning via Retrying","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Paavo Parmas, Shin Ishii, Soichiro Nishimori, Sotetsu Koyamada, Tadashi Kozuno, Toshinori Kitamura, Yutaka Matsuo","submitted_at":"2026-05-29T03:35:13Z","abstract_excerpt":"In reinforcement learning (RL), agents benefit from exploration only because they repeatedly encounter similar states: trying different actions can improve performance or reduce uncertainty; without such retries, a greedy policy is optimal. We formalize this intuition with ReMax, an objective that evaluates a policy by the expected maximum return over $M$ samples, where $M$ is a positive integer, while accounting for return uncertainty. Optimizing this objective induces stochastic exploration as an emergent property, without explicit bonus terms. For efficient policy optimization, we derive a "},"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":"2606.00151","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T03:35:13Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3328269df6096d1f71be26d4fb99a69220561d38072160173ab2aa701eec29f4","abstract_canon_sha256":"a2ae74dc996c7f04d04976662c7975bae6bd04d9c9b7a259747bc0ae4a50b7d9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:19.894141Z","signature_b64":"F93mnCoMiyQnBCyT/TL43y0mYeUwcA4a6Uce+50B20Ys3KFyj5PLHtiYmD9EjRjb7v+qVJ3jsBMaYRHn0O2+AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0479365824dff8fb6b2ed8992a91006f79e5a3c9b946a0616855ee25b842ffdb","last_reissued_at":"2026-06-02T01:03:19.893727Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:19.893727Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Emergence of Exploration in Policy Gradient Reinforcement Learning via Retrying","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Paavo Parmas, Shin Ishii, Soichiro Nishimori, Sotetsu Koyamada, Tadashi Kozuno, Toshinori Kitamura, Yutaka Matsuo","submitted_at":"2026-05-29T03:35:13Z","abstract_excerpt":"In reinforcement learning (RL), agents benefit from exploration only because they repeatedly encounter similar states: trying different actions can improve performance or reduce uncertainty; without such retries, a greedy policy is optimal. We formalize this intuition with ReMax, an objective that evaluates a policy by the expected maximum return over $M$ samples, where $M$ is a positive integer, while accounting for return uncertainty. Optimizing this objective induces stochastic exploration as an emergent property, without explicit bonus terms. For efficient policy optimization, we derive a "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.00151","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/2606.00151/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":"2606.00151","created_at":"2026-06-02T01:03:19.893786+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.00151v1","created_at":"2026-06-02T01:03:19.893786+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.00151","created_at":"2026-06-02T01:03:19.893786+00:00"},{"alias_kind":"pith_short_12","alias_value":"AR4TMWBE374P","created_at":"2026-06-02T01:03:19.893786+00:00"},{"alias_kind":"pith_short_16","alias_value":"AR4TMWBE374PW2ZO","created_at":"2026-06-02T01:03:19.893786+00:00"},{"alias_kind":"pith_short_8","alias_value":"AR4TMWBE","created_at":"2026-06-02T01:03:19.893786+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5","json":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5.json","graph_json":"https://pith.science/api/pith-number/AR4TMWBE374PW2ZO3CMSVEIAN5/graph.json","events_json":"https://pith.science/api/pith-number/AR4TMWBE374PW2ZO3CMSVEIAN5/events.json","paper":"https://pith.science/paper/AR4TMWBE"},"agent_actions":{"view_html":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5","download_json":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5.json","view_paper":"https://pith.science/paper/AR4TMWBE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.00151&json=true","fetch_graph":"https://pith.science/api/pith-number/AR4TMWBE374PW2ZO3CMSVEIAN5/graph.json","fetch_events":"https://pith.science/api/pith-number/AR4TMWBE374PW2ZO3CMSVEIAN5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5/action/storage_attestation","attest_author":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5/action/author_attestation","sign_citation":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5/action/citation_signature","submit_replication":"https://pith.science/pith/AR4TMWBE374PW2ZO3CMSVEIAN5/action/replication_record"}},"created_at":"2026-06-02T01:03:19.893786+00:00","updated_at":"2026-06-02T01:03:19.893786+00:00"}