{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:EOAROZ6QDZXXRBVGEIHPARTKSC","short_pith_number":"pith:EOAROZ6Q","schema_version":"1.0","canonical_sha256":"23811767d01e6f7886a6220ef0466a909fc1124c404beea583099a265e092985","source":{"kind":"arxiv","id":"1805.11447","version":1},"attestation_state":"computed","paper":{"title":"Virtuously Safe Reinforcement Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.GT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alexandre Maurer, El Mahdi El Mhamdi, Henrik Aslund, Rachid Guerraoui","submitted_at":"2018-05-29T13:34:39Z","abstract_excerpt":"We show that when a third party, the adversary, steps into the two-party setting (agent and operator) of safely interruptible reinforcement learning, a trade-off has to be made between the probability of following the optimal policy in the limit, and the probability of escaping a dangerous situation created by the adversary. So far, the work on safely interruptible agents has assumed a perfect perception of the agent about its environment (no adversary), and therefore implicitly set the second probability to zero, by explicitly seeking a value of one for the first probability. We show that (1)"},"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":"1805.11447","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2018-05-29T13:34:39Z","cross_cats_sorted":["cs.AI","cs.GT","stat.ML"],"title_canon_sha256":"1a46a5ade1d0480c033da7adf4a072281f47b507bd3a204fff100620da7e3c0f","abstract_canon_sha256":"2278822d512fe29fc7ee54f76880a9321c007fef2cd2499a59e117f7182aaa16"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:14:42.528736Z","signature_b64":"89AG303iJ/fU+UNlKW6pA5R+DneBxLpriAwpesCLjVKh+4g9sLTKZBG8zGee95GQO0HLpRQzS66C6UwAzeLzCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"23811767d01e6f7886a6220ef0466a909fc1124c404beea583099a265e092985","last_reissued_at":"2026-05-18T00:14:42.528201Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:14:42.528201Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Virtuously Safe Reinforcement Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.GT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alexandre Maurer, El Mahdi El Mhamdi, Henrik Aslund, Rachid Guerraoui","submitted_at":"2018-05-29T13:34:39Z","abstract_excerpt":"We show that when a third party, the adversary, steps into the two-party setting (agent and operator) of safely interruptible reinforcement learning, a trade-off has to be made between the probability of following the optimal policy in the limit, and the probability of escaping a dangerous situation created by the adversary. So far, the work on safely interruptible agents has assumed a perfect perception of the agent about its environment (no adversary), and therefore implicitly set the second probability to zero, by explicitly seeking a value of one for the first probability. We show that (1)"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1805.11447","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":""},"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":"1805.11447","created_at":"2026-05-18T00:14:42.528289+00:00"},{"alias_kind":"arxiv_version","alias_value":"1805.11447v1","created_at":"2026-05-18T00:14:42.528289+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1805.11447","created_at":"2026-05-18T00:14:42.528289+00:00"},{"alias_kind":"pith_short_12","alias_value":"EOAROZ6QDZXX","created_at":"2026-05-18T12:32:22.470017+00:00"},{"alias_kind":"pith_short_16","alias_value":"EOAROZ6QDZXXRBVG","created_at":"2026-05-18T12:32:22.470017+00:00"},{"alias_kind":"pith_short_8","alias_value":"EOAROZ6Q","created_at":"2026-05-18T12:32:22.470017+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2603.16651","citing_title":"What if Pinocchio Were a Reinforcement Learning Agent: A Normative End-to-End Pipeline","ref_index":29,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC","json":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC.json","graph_json":"https://pith.science/api/pith-number/EOAROZ6QDZXXRBVGEIHPARTKSC/graph.json","events_json":"https://pith.science/api/pith-number/EOAROZ6QDZXXRBVGEIHPARTKSC/events.json","paper":"https://pith.science/paper/EOAROZ6Q"},"agent_actions":{"view_html":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC","download_json":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC.json","view_paper":"https://pith.science/paper/EOAROZ6Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1805.11447&json=true","fetch_graph":"https://pith.science/api/pith-number/EOAROZ6QDZXXRBVGEIHPARTKSC/graph.json","fetch_events":"https://pith.science/api/pith-number/EOAROZ6QDZXXRBVGEIHPARTKSC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC/action/storage_attestation","attest_author":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC/action/author_attestation","sign_citation":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC/action/citation_signature","submit_replication":"https://pith.science/pith/EOAROZ6QDZXXRBVGEIHPARTKSC/action/replication_record"}},"created_at":"2026-05-18T00:14:42.528289+00:00","updated_at":"2026-05-18T00:14:42.528289+00:00"}