{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AKT4PA27VFE5BJD7YPCZ75ZTEF","short_pith_number":"pith:AKT4PA27","schema_version":"1.0","canonical_sha256":"02a7c7835fa949d0a47fc3c59ff73321548bc7f5ee6ec7f71478e8fe76a26526","source":{"kind":"arxiv","id":"2605.11975","version":2},"attestation_state":"computed","paper":{"title":"Stochastic Minimum-Cost Reach-Avoid Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Reach-avoid probability certificates turn stochastic safety constraints into a surrogate objective that reinforcement learners can optimize for minimum cost.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bai Xue, Jingduo Pan, Taoran Wu, Yiling Xue","submitted_at":"2026-05-12T11:31:36Z","abstract_excerpt":"We study stochastic minimum-cost reach-avoid reinforcement learning, where an agent must satisfy a reach-avoid specification with probability at least $p$ while minimizing expected cumulative costs in stochastic environments. Existing safe and constrained reinforcement learning methods typically fail to jointly enforce probabilistic reach-avoid constraints and optimize cost in the learning setting in stochastic environments. To address this challenge, we introduce reach-avoid probability certificates (RAPCs), which identify states from which stochastic reach-avoid constraints are satisfiable. "},"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":true},"canonical_record":{"source":{"id":"2605.11975","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T11:31:36Z","cross_cats_sorted":[],"title_canon_sha256":"66d720cff133972aa7823b5ebaac80077487d3c568a5519b8ddfdb598a6eb03a","abstract_canon_sha256":"0a8a6fcc79b1dafa5a3fd8818da64a1986023d846c8114d99a5698cd87cbac34"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:36.371184Z","signature_b64":"WRqpCkg+0T2fhlfdtDikkWkvvI6Ioz9Sb/Yx5sMSylWGTPipgLkkFamoAHVzCaWltmYCkoT+ps+OsoNUWppXDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"02a7c7835fa949d0a47fc3c59ff73321548bc7f5ee6ec7f71478e8fe76a26526","last_reissued_at":"2026-05-20T00:04:36.370095Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:36.370095Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stochastic Minimum-Cost Reach-Avoid Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Reach-avoid probability certificates turn stochastic safety constraints into a surrogate objective that reinforcement learners can optimize for minimum cost.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bai Xue, Jingduo Pan, Taoran Wu, Yiling Xue","submitted_at":"2026-05-12T11:31:36Z","abstract_excerpt":"We study stochastic minimum-cost reach-avoid reinforcement learning, where an agent must satisfy a reach-avoid specification with probability at least $p$ while minimizing expected cumulative costs in stochastic environments. Existing safe and constrained reinforcement learning methods typically fail to jointly enforce probabilistic reach-avoid constraints and optimize cost in the learning setting in stochastic environments. To address this challenge, we introduce reach-avoid probability certificates (RAPCs), which identify states from which stochastic reach-avoid constraints are satisfiable. "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We establish almost sure convergence of the proposed algorithms to locally optimal policies with respect to the resulting objective.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That reach-avoid probability certificates can be computed or approximated accurately enough during learning to serve as a reliable surrogate for the true probabilistic constraint in stochastic environments.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Introduces RAPCs and a contraction Bellman operator that jointly enforce probabilistic reach-avoid constraints while minimizing expected costs in stochastic RL, with almost-sure convergence to local optima.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reach-avoid probability certificates turn stochastic safety constraints into a surrogate objective that reinforcement learners can optimize for minimum cost.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"cb98d4fcf3484c15be6e8aa3859b7521c610e76c5361292e128e68ddb09613eb"},"source":{"id":"2605.11975","kind":"arxiv","version":2},"verdict":{"id":"dbc54e43-26df-498b-8077-da1985984ccf","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T07:23:59.690351Z","strongest_claim":"We establish almost sure convergence of the proposed algorithms to locally optimal policies with respect to the resulting objective.","one_line_summary":"Introduces RAPCs and a contraction Bellman operator that jointly enforce probabilistic reach-avoid constraints while minimizing expected costs in stochastic RL, with almost-sure convergence to local optima.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That reach-avoid probability certificates can be computed or approximated accurately enough during learning to serve as a reliable surrogate for the true probabilistic constraint in stochastic environments.","pith_extraction_headline":"Reach-avoid probability certificates turn stochastic safety constraints into a surrogate objective that reinforcement learners can optimize for minimum cost."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.11975/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T11:34:51.555285Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T09:01:17.174166Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T08:02:38.043163Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"64248693b94f85b78bf902b8a5cb22190e5b51432067cb8d1578a25192c7eb03"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b89d607eea2972b64737389308f812ef68667cb70e74931bfba336853027a791"},"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":"2605.11975","created_at":"2026-05-20T00:04:36.370252+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.11975v2","created_at":"2026-05-20T00:04:36.370252+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.11975","created_at":"2026-05-20T00:04:36.370252+00:00"},{"alias_kind":"pith_short_12","alias_value":"AKT4PA27VFE5","created_at":"2026-05-20T00:04:36.370252+00:00"},{"alias_kind":"pith_short_16","alias_value":"AKT4PA27VFE5BJD7","created_at":"2026-05-20T00:04:36.370252+00:00"},{"alias_kind":"pith_short_8","alias_value":"AKT4PA27","created_at":"2026-05-20T00:04:36.370252+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF","json":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF.json","graph_json":"https://pith.science/api/pith-number/AKT4PA27VFE5BJD7YPCZ75ZTEF/graph.json","events_json":"https://pith.science/api/pith-number/AKT4PA27VFE5BJD7YPCZ75ZTEF/events.json","paper":"https://pith.science/paper/AKT4PA27"},"agent_actions":{"view_html":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF","download_json":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF.json","view_paper":"https://pith.science/paper/AKT4PA27","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.11975&json=true","fetch_graph":"https://pith.science/api/pith-number/AKT4PA27VFE5BJD7YPCZ75ZTEF/graph.json","fetch_events":"https://pith.science/api/pith-number/AKT4PA27VFE5BJD7YPCZ75ZTEF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF/action/storage_attestation","attest_author":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF/action/author_attestation","sign_citation":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF/action/citation_signature","submit_replication":"https://pith.science/pith/AKT4PA27VFE5BJD7YPCZ75ZTEF/action/replication_record"}},"created_at":"2026-05-20T00:04:36.370252+00:00","updated_at":"2026-05-20T00:04:36.370252+00:00"}