{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IXKXNWD3NCEOIMEMRFWQ25QHD6","short_pith_number":"pith:IXKXNWD3","schema_version":"1.0","canonical_sha256":"45d576d87b6888e4308c896d0d76071fb3b8fb0f75e71da9f16c805d7ec4b99e","source":{"kind":"arxiv","id":"2602.04599","version":2},"attestation_state":"computed","paper":{"title":"Stochastic Decision Horizons for Constrained Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Daniel Haeufle, Georg Martius, Leonard Franz, Nico Scherf, Nikola Milosevic, Pavel Kolev","submitted_at":"2026-02-04T14:27:16Z","abstract_excerpt":"We propose stochastic decision horizons (SDH), a theoretically grounded framework for solving constrained RL problems with every-step constraint satisfaction, a desirable property in many real-world applications. In SDH, a constraint violation yields an effective shortening of horizon via a state-action continuation probability. Using Control as Inference, we develop the first off-policy and regularized algorithms for RL with instantaneous constraints. We identify two principled semantics for what counts as a decision after a violation. Absorbing-state semantics end the decision process, so on"},"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":"2602.04599","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-04T14:27:16Z","cross_cats_sorted":[],"title_canon_sha256":"73263b68126cd4e8a12136f246531a85318183bf6b5eadb62d9d0d2cfeac905c","abstract_canon_sha256":"5af913a1cf9db9c60178d43950087d21cb30420ef2d5f5a2c052b55f030426bd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:05:44.025748Z","signature_b64":"fX1rxbTQvQQD+JEvQYtYkZRJ8iHouuOTrsk+/UePr6DcNGqAR4dehVygl6mzGQHnBLElpmusr0qgsnKTOJRIAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45d576d87b6888e4308c896d0d76071fb3b8fb0f75e71da9f16c805d7ec4b99e","last_reissued_at":"2026-05-27T01:05:44.024784Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:05:44.024784Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stochastic Decision Horizons for Constrained Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Daniel Haeufle, Georg Martius, Leonard Franz, Nico Scherf, Nikola Milosevic, Pavel Kolev","submitted_at":"2026-02-04T14:27:16Z","abstract_excerpt":"We propose stochastic decision horizons (SDH), a theoretically grounded framework for solving constrained RL problems with every-step constraint satisfaction, a desirable property in many real-world applications. In SDH, a constraint violation yields an effective shortening of horizon via a state-action continuation probability. Using Control as Inference, we develop the first off-policy and regularized algorithms for RL with instantaneous constraints. We identify two principled semantics for what counts as a decision after a violation. Absorbing-state semantics end the decision process, so on"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.04599","kind":"arxiv","version":2},"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/2602.04599/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":"2602.04599","created_at":"2026-05-27T01:05:44.024911+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.04599v2","created_at":"2026-05-27T01:05:44.024911+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.04599","created_at":"2026-05-27T01:05:44.024911+00:00"},{"alias_kind":"pith_short_12","alias_value":"IXKXNWD3NCEO","created_at":"2026-05-27T01:05:44.024911+00:00"},{"alias_kind":"pith_short_16","alias_value":"IXKXNWD3NCEOIMEM","created_at":"2026-05-27T01:05:44.024911+00:00"},{"alias_kind":"pith_short_8","alias_value":"IXKXNWD3","created_at":"2026-05-27T01:05:44.024911+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/IXKXNWD3NCEOIMEMRFWQ25QHD6","json":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6.json","graph_json":"https://pith.science/api/pith-number/IXKXNWD3NCEOIMEMRFWQ25QHD6/graph.json","events_json":"https://pith.science/api/pith-number/IXKXNWD3NCEOIMEMRFWQ25QHD6/events.json","paper":"https://pith.science/paper/IXKXNWD3"},"agent_actions":{"view_html":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6","download_json":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6.json","view_paper":"https://pith.science/paper/IXKXNWD3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.04599&json=true","fetch_graph":"https://pith.science/api/pith-number/IXKXNWD3NCEOIMEMRFWQ25QHD6/graph.json","fetch_events":"https://pith.science/api/pith-number/IXKXNWD3NCEOIMEMRFWQ25QHD6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6/action/storage_attestation","attest_author":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6/action/author_attestation","sign_citation":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6/action/citation_signature","submit_replication":"https://pith.science/pith/IXKXNWD3NCEOIMEMRFWQ25QHD6/action/replication_record"}},"created_at":"2026-05-27T01:05:44.024911+00:00","updated_at":"2026-05-27T01:05:44.024911+00:00"}