{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CLKGG7GH3AZD7NMO22XQLBJ7OR","short_pith_number":"pith:CLKGG7GH","schema_version":"1.0","canonical_sha256":"12d4637cc7d8323fb58ed6af05853f747e55a2b67d4353530956dd25adbf10a5","source":{"kind":"arxiv","id":"2601.22993","version":4},"attestation_state":"computed","paper":{"title":"Constrained Policy Optimization with Cantelli-Bounded Value-at-Risk","license":"http://creativecommons.org/licenses/by/4.0/","headline":"VaR-CPO approximates Value-at-Risk constraints via Cantelli's inequality to guarantee zero violations during training in feasible reinforcement learning environments.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Jan-Peter Calliess, Rohan Tangri","submitted_at":"2026-01-30T13:57:47Z","abstract_excerpt":"We introduce Canary, a risk-averse method designed to optimize Value-at-Risk (VaR) constrained reinforcement learning (RL) problems. We employ Cantelli's inequality to obtain a tractable, conservative and smooth bound on the VaR constraint based on the first two moments of the cost return. This yields a constraint estimator that remains stable with tight violation thresholds in dense cost regimes. Extending the trust-region framework of the Constrained Policy Optimization (CPO) method, we further provide worst-case bounds for both policy improvement and constraint violation during the training"},"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":"2601.22993","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-30T13:57:47Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"30ce4acde7984652dff459c97f283d90840781c4c1a41d0540b94b58f3c4c83c","abstract_canon_sha256":"a662487e501e5952911af607c6e10b4bdbd3b18692871938a12ba80531e99d48"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:17:33.334599Z","signature_b64":"zRk+JsQwltP0gDPfrzWW5Rck7jc6oyIh9IvsqCG5BMtjvwuYhucVBTnKWkna3PC9Z7oHB6cbwaJFsIXBy+2pDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"12d4637cc7d8323fb58ed6af05853f747e55a2b67d4353530956dd25adbf10a5","last_reissued_at":"2026-06-30T01:17:33.333863Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:17:33.333863Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Constrained Policy Optimization with Cantelli-Bounded Value-at-Risk","license":"http://creativecommons.org/licenses/by/4.0/","headline":"VaR-CPO approximates Value-at-Risk constraints via Cantelli's inequality to guarantee zero violations during training in feasible reinforcement learning environments.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Jan-Peter Calliess, Rohan Tangri","submitted_at":"2026-01-30T13:57:47Z","abstract_excerpt":"We introduce Canary, a risk-averse method designed to optimize Value-at-Risk (VaR) constrained reinforcement learning (RL) problems. We employ Cantelli's inequality to obtain a tractable, conservative and smooth bound on the VaR constraint based on the first two moments of the cost return. This yields a constraint estimator that remains stable with tight violation thresholds in dense cost regimes. Extending the trust-region framework of the Constrained Policy Optimization (CPO) method, we further provide worst-case bounds for both policy improvement and constraint violation during the training"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"VaR-CPO achieves zero constraint violations during training in feasible environments, a property baseline methods fail to uphold, while providing worst-case bounds for policy improvement and constraint violation.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The Cantelli inequality provides a sufficiently tight and conservative approximation to the true VaR constraint so that enforcing the bound still guarantees the original probabilistic constraint in practice.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"VaR-CPO approximates non-differentiable VaR constraints via Cantelli's inequality to enable safe, sample-efficient policy optimization with zero training violations in feasible environments.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"VaR-CPO approximates Value-at-Risk constraints via Cantelli's inequality to guarantee zero violations during training in feasible reinforcement learning environments.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5833deae7b8bf68db5033edbd4f3443af94e8f792a74de4f7a049cff80bd6916"},"source":{"id":"2601.22993","kind":"arxiv","version":4},"verdict":{"id":"7b84fecb-eea6-421d-90b3-a8385299ea5b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T09:35:17.857407Z","strongest_claim":"VaR-CPO achieves zero constraint violations during training in feasible environments, a property baseline methods fail to uphold, while providing worst-case bounds for policy improvement and constraint violation.","one_line_summary":"VaR-CPO approximates non-differentiable VaR constraints via Cantelli's inequality to enable safe, sample-efficient policy optimization with zero training violations in feasible environments.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The Cantelli inequality provides a sufficiently tight and conservative approximation to the true VaR constraint so that enforcing the bound still guarantees the original probabilistic constraint in practice.","pith_extraction_headline":"VaR-CPO approximates Value-at-Risk constraints via Cantelli's inequality to guarantee zero violations during training in feasible reinforcement learning environments."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2601.22993/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":2,"snapshot_sha256":"9040767e148b8c72e9fc099919b372291776c02fefffa41b70a4c425745a302e"},"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":"2601.22993","created_at":"2026-06-30T01:17:33.333949+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.22993v4","created_at":"2026-06-30T01:17:33.333949+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.22993","created_at":"2026-06-30T01:17:33.333949+00:00"},{"alias_kind":"pith_short_12","alias_value":"CLKGG7GH3AZD","created_at":"2026-06-30T01:17:33.333949+00:00"},{"alias_kind":"pith_short_16","alias_value":"CLKGG7GH3AZD7NMO","created_at":"2026-06-30T01:17:33.333949+00:00"},{"alias_kind":"pith_short_8","alias_value":"CLKGG7GH","created_at":"2026-06-30T01:17:33.333949+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/CLKGG7GH3AZD7NMO22XQLBJ7OR","json":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR.json","graph_json":"https://pith.science/api/pith-number/CLKGG7GH3AZD7NMO22XQLBJ7OR/graph.json","events_json":"https://pith.science/api/pith-number/CLKGG7GH3AZD7NMO22XQLBJ7OR/events.json","paper":"https://pith.science/paper/CLKGG7GH"},"agent_actions":{"view_html":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR","download_json":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR.json","view_paper":"https://pith.science/paper/CLKGG7GH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.22993&json=true","fetch_graph":"https://pith.science/api/pith-number/CLKGG7GH3AZD7NMO22XQLBJ7OR/graph.json","fetch_events":"https://pith.science/api/pith-number/CLKGG7GH3AZD7NMO22XQLBJ7OR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR/action/storage_attestation","attest_author":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR/action/author_attestation","sign_citation":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR/action/citation_signature","submit_replication":"https://pith.science/pith/CLKGG7GH3AZD7NMO22XQLBJ7OR/action/replication_record"}},"created_at":"2026-06-30T01:17:33.333949+00:00","updated_at":"2026-06-30T01:17:33.333949+00:00"}