{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IUCZFJFH5K4XJVTLCD2WHQ3LCA","short_pith_number":"pith:IUCZFJFH","schema_version":"1.0","canonical_sha256":"450592a4a7eab974d66b10f563c36b102445a31d34a504e527b7ad746e07e150","source":{"kind":"arxiv","id":"2507.14901","version":1},"attestation_state":"computed","paper":{"title":"Learning Nonlinear Causal Reductions to Explain Reinforcement Learning Policies","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Armin Keki\\'c, Bernhard Sch\\\"olkopf, Dieter B\\\"uchler, Jan Schneider, Michel Besserve","submitted_at":"2025-07-20T10:25:24Z","abstract_excerpt":"Why do reinforcement learning (RL) policies fail or succeed? This is a challenging question due to the complex, high-dimensional nature of agent-environment interactions. In this work, we take a causal perspective on explaining the behavior of RL policies by viewing the states, actions, and rewards as variables in a low-level causal model. We introduce random perturbations to policy actions during execution and observe their effects on the cumulative reward, learning a simplified high-level causal model that explains these relationships. To this end, we develop a nonlinear Causal Model Reducti"},"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":"2507.14901","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-07-20T10:25:24Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"58780974594671a77a8397c4c2dd7b8e2edc5585440c04fb3984aaa9799cafc5","abstract_canon_sha256":"6565a29cc3131569fe4bcb58e6fdc965191db790855686f40eb1bf7bc35ffa1e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:40:15.307715Z","signature_b64":"/AbPWaIFL1gN22LPovcFT6UFWM3YFcSmCu/EGZ6MebUwWSGpTesNlZXA3gE2SMWidAXbq163oLdbO5QXQxgbAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"450592a4a7eab974d66b10f563c36b102445a31d34a504e527b7ad746e07e150","last_reissued_at":"2026-07-05T11:40:15.307243Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:40:15.307243Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Nonlinear Causal Reductions to Explain Reinforcement Learning Policies","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Armin Keki\\'c, Bernhard Sch\\\"olkopf, Dieter B\\\"uchler, Jan Schneider, Michel Besserve","submitted_at":"2025-07-20T10:25:24Z","abstract_excerpt":"Why do reinforcement learning (RL) policies fail or succeed? This is a challenging question due to the complex, high-dimensional nature of agent-environment interactions. In this work, we take a causal perspective on explaining the behavior of RL policies by viewing the states, actions, and rewards as variables in a low-level causal model. We introduce random perturbations to policy actions during execution and observe their effects on the cumulative reward, learning a simplified high-level causal model that explains these relationships. To this end, we develop a nonlinear Causal Model Reducti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.14901","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/2507.14901/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":"2507.14901","created_at":"2026-07-05T11:40:15.307299+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.14901v1","created_at":"2026-07-05T11:40:15.307299+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.14901","created_at":"2026-07-05T11:40:15.307299+00:00"},{"alias_kind":"pith_short_12","alias_value":"IUCZFJFH5K4X","created_at":"2026-07-05T11:40:15.307299+00:00"},{"alias_kind":"pith_short_16","alias_value":"IUCZFJFH5K4XJVTL","created_at":"2026-07-05T11:40:15.307299+00:00"},{"alias_kind":"pith_short_8","alias_value":"IUCZFJFH","created_at":"2026-07-05T11:40:15.307299+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18089","citing_title":"From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning","ref_index":100,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA","json":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA.json","graph_json":"https://pith.science/api/pith-number/IUCZFJFH5K4XJVTLCD2WHQ3LCA/graph.json","events_json":"https://pith.science/api/pith-number/IUCZFJFH5K4XJVTLCD2WHQ3LCA/events.json","paper":"https://pith.science/paper/IUCZFJFH"},"agent_actions":{"view_html":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA","download_json":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA.json","view_paper":"https://pith.science/paper/IUCZFJFH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.14901&json=true","fetch_graph":"https://pith.science/api/pith-number/IUCZFJFH5K4XJVTLCD2WHQ3LCA/graph.json","fetch_events":"https://pith.science/api/pith-number/IUCZFJFH5K4XJVTLCD2WHQ3LCA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA/action/storage_attestation","attest_author":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA/action/author_attestation","sign_citation":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA/action/citation_signature","submit_replication":"https://pith.science/pith/IUCZFJFH5K4XJVTLCD2WHQ3LCA/action/replication_record"}},"created_at":"2026-07-05T11:40:15.307299+00:00","updated_at":"2026-07-05T11:40:15.307299+00:00"}