{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:L3SPIZASKIYU2RGDQ3LG3L6QRD","short_pith_number":"pith:L3SPIZAS","schema_version":"1.0","canonical_sha256":"5ee4f4641252314d44c386d66dafd088dac03827e6016c6c5578698298f29a4a","source":{"kind":"arxiv","id":"2112.06625","version":1},"attestation_state":"computed","paper":{"title":"Lifelong Hyper-Policy Optimization with Multiple Importance Sampling Regularization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alberto Maria Metelli, Francesco Vidaich, Marcello Restelli, Pierre Liotet","submitted_at":"2021-12-13T13:09:49Z","abstract_excerpt":"Learning in a lifelong setting, where the dynamics continually evolve, is a hard challenge for current reinforcement learning algorithms. Yet this would be a much needed feature for practical applications. In this paper, we propose an approach which learns a hyper-policy, whose input is time, that outputs the parameters of the policy to be queried at that time. This hyper-policy is trained to maximize the estimated future performance, efficiently reusing past data by means of importance sampling, at the cost of introducing a controlled bias. We combine the future performance estimate with the "},"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":"2112.06625","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-12-13T13:09:49Z","cross_cats_sorted":[],"title_canon_sha256":"681bde8c973b0dacae8d194b6cc8677852a1ac2ca3576b7bdf1a4a6083258278","abstract_canon_sha256":"5f0e6aaece0ec7e47c333e5694f530a1cf51ea561782214cb5c3ef4ba44a0eb1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:40:15.508761Z","signature_b64":"yb11xeSV2DTPmhi6FNvcFx803YA7M2zm4q/6VOmZVJp+J36gVIdZGSwGoSc3fSJ+t1JKP4zZKKn/IzN2t1KrDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5ee4f4641252314d44c386d66dafd088dac03827e6016c6c5578698298f29a4a","last_reissued_at":"2026-07-05T03:40:15.508351Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:40:15.508351Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Lifelong Hyper-Policy Optimization with Multiple Importance Sampling Regularization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alberto Maria Metelli, Francesco Vidaich, Marcello Restelli, Pierre Liotet","submitted_at":"2021-12-13T13:09:49Z","abstract_excerpt":"Learning in a lifelong setting, where the dynamics continually evolve, is a hard challenge for current reinforcement learning algorithms. Yet this would be a much needed feature for practical applications. In this paper, we propose an approach which learns a hyper-policy, whose input is time, that outputs the parameters of the policy to be queried at that time. This hyper-policy is trained to maximize the estimated future performance, efficiently reusing past data by means of importance sampling, at the cost of introducing a controlled bias. We combine the future performance estimate with the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2112.06625","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/2112.06625/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":"2112.06625","created_at":"2026-07-05T03:40:15.508408+00:00"},{"alias_kind":"arxiv_version","alias_value":"2112.06625v1","created_at":"2026-07-05T03:40:15.508408+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2112.06625","created_at":"2026-07-05T03:40:15.508408+00:00"},{"alias_kind":"pith_short_12","alias_value":"L3SPIZASKIYU","created_at":"2026-07-05T03:40:15.508408+00:00"},{"alias_kind":"pith_short_16","alias_value":"L3SPIZASKIYU2RGD","created_at":"2026-07-05T03:40:15.508408+00:00"},{"alias_kind":"pith_short_8","alias_value":"L3SPIZAS","created_at":"2026-07-05T03:40:15.508408+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/L3SPIZASKIYU2RGDQ3LG3L6QRD","json":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD.json","graph_json":"https://pith.science/api/pith-number/L3SPIZASKIYU2RGDQ3LG3L6QRD/graph.json","events_json":"https://pith.science/api/pith-number/L3SPIZASKIYU2RGDQ3LG3L6QRD/events.json","paper":"https://pith.science/paper/L3SPIZAS"},"agent_actions":{"view_html":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD","download_json":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD.json","view_paper":"https://pith.science/paper/L3SPIZAS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2112.06625&json=true","fetch_graph":"https://pith.science/api/pith-number/L3SPIZASKIYU2RGDQ3LG3L6QRD/graph.json","fetch_events":"https://pith.science/api/pith-number/L3SPIZASKIYU2RGDQ3LG3L6QRD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD/action/storage_attestation","attest_author":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD/action/author_attestation","sign_citation":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD/action/citation_signature","submit_replication":"https://pith.science/pith/L3SPIZASKIYU2RGDQ3LG3L6QRD/action/replication_record"}},"created_at":"2026-07-05T03:40:15.508408+00:00","updated_at":"2026-07-05T03:40:15.508408+00:00"}