{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:63F6IZ2QFSL6Q5BGYMEHV7H3X3","short_pith_number":"pith:63F6IZ2Q","schema_version":"1.0","canonical_sha256":"f6cbe467502c97e87426c3087afcfbbec684164dc992915b09dd7b34c9d3fa7e","source":{"kind":"arxiv","id":"2504.16879","version":1},"attestation_state":"computed","paper":{"title":"Learning Verifiable Control Policies Using Relaxed Verification","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SY"],"primary_cat":"eess.SY","authors_text":"Alexander Estornell, Michael Everett, Puja Chaudhury","submitted_at":"2025-04-23T16:54:35Z","abstract_excerpt":"To provide safety guarantees for learning-based control systems, recent work has developed formal verification methods to apply after training ends. However, if the trained policy does not meet the specifications, or there is conservatism in the verification algorithm, establishing these guarantees may not be possible. Instead, this work proposes to perform verification throughout training to ultimately aim for policies whose properties can be evaluated throughout runtime with lightweight, relaxed verification algorithms. The approach is to use differentiable reachability analysis and incorpor"},"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":"2504.16879","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.SY","submitted_at":"2025-04-23T16:54:35Z","cross_cats_sorted":["cs.LG","cs.SY"],"title_canon_sha256":"1caa9e7a45e6dc31fdd82b4251bb665bea5c368649d66360620e03b21bcf2548","abstract_canon_sha256":"bc72cad1e272cd90dd429ebfe7ed97318a7fde3250f460f16428f9e1e6e595c8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:53:05.798615Z","signature_b64":"alIdN74hYwwLVD4ToyOmm2G11fuf8Avuu1ItcAjinmI5G12i7mwYiAQFolxQQO9tPJmYEn6wGd1Qm2gA1XIoAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f6cbe467502c97e87426c3087afcfbbec684164dc992915b09dd7b34c9d3fa7e","last_reissued_at":"2026-07-05T10:53:05.798078Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:53:05.798078Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Verifiable Control Policies Using Relaxed Verification","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SY"],"primary_cat":"eess.SY","authors_text":"Alexander Estornell, Michael Everett, Puja Chaudhury","submitted_at":"2025-04-23T16:54:35Z","abstract_excerpt":"To provide safety guarantees for learning-based control systems, recent work has developed formal verification methods to apply after training ends. However, if the trained policy does not meet the specifications, or there is conservatism in the verification algorithm, establishing these guarantees may not be possible. Instead, this work proposes to perform verification throughout training to ultimately aim for policies whose properties can be evaluated throughout runtime with lightweight, relaxed verification algorithms. The approach is to use differentiable reachability analysis and incorpor"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.16879","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/2504.16879/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":"2504.16879","created_at":"2026-07-05T10:53:05.798139+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.16879v1","created_at":"2026-07-05T10:53:05.798139+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.16879","created_at":"2026-07-05T10:53:05.798139+00:00"},{"alias_kind":"pith_short_12","alias_value":"63F6IZ2QFSL6","created_at":"2026-07-05T10:53:05.798139+00:00"},{"alias_kind":"pith_short_16","alias_value":"63F6IZ2QFSL6Q5BG","created_at":"2026-07-05T10:53:05.798139+00:00"},{"alias_kind":"pith_short_8","alias_value":"63F6IZ2Q","created_at":"2026-07-05T10:53:05.798139+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.30935","citing_title":"ShardNet: Training Neural Controllers with Hard, Non-Convex Constraints","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3","json":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3.json","graph_json":"https://pith.science/api/pith-number/63F6IZ2QFSL6Q5BGYMEHV7H3X3/graph.json","events_json":"https://pith.science/api/pith-number/63F6IZ2QFSL6Q5BGYMEHV7H3X3/events.json","paper":"https://pith.science/paper/63F6IZ2Q"},"agent_actions":{"view_html":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3","download_json":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3.json","view_paper":"https://pith.science/paper/63F6IZ2Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.16879&json=true","fetch_graph":"https://pith.science/api/pith-number/63F6IZ2QFSL6Q5BGYMEHV7H3X3/graph.json","fetch_events":"https://pith.science/api/pith-number/63F6IZ2QFSL6Q5BGYMEHV7H3X3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3/action/storage_attestation","attest_author":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3/action/author_attestation","sign_citation":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3/action/citation_signature","submit_replication":"https://pith.science/pith/63F6IZ2QFSL6Q5BGYMEHV7H3X3/action/replication_record"}},"created_at":"2026-07-05T10:53:05.798139+00:00","updated_at":"2026-07-05T10:53:05.798139+00:00"}