{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:WKJBMHO7RQLLX7NVPR4D5P3N42","short_pith_number":"pith:WKJBMHO7","schema_version":"1.0","canonical_sha256":"b292161ddf8c16bbfdb57c783ebf6de6b96b105467b76f481300ed245d84a95f","source":{"kind":"arxiv","id":"2206.01367","version":1},"attestation_state":"computed","paper":{"title":"Adversarial Unlearning: Reducing Confidence Along Adversarial Directions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Amrith Setlur, Benjamin Eysenbach, Sergey Levine, Virginia Smith","submitted_at":"2022-06-03T02:26:24Z","abstract_excerpt":"Supervised learning methods trained with maximum likelihood objectives often overfit on training data. Most regularizers that prevent overfitting look to increase confidence on additional examples (e.g., data augmentation, adversarial training), or reduce it on training data (e.g., label smoothing). In this work we propose a complementary regularization strategy that reduces confidence on self-generated examples. The method, which we call RCAD (Reducing Confidence along Adversarial Directions), aims to reduce confidence on out-of-distribution examples lying along directions adversarially chose"},"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":"2206.01367","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-06-03T02:26:24Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"c2e0b731f20b2bbb69a11bca8ea569ab4d3d7aee3d8bc249439977f4640a3d26","abstract_canon_sha256":"436c88971e54be5ed3e9760957f810b31bb6acc6341dae5b92f42e9b4cebb157"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:28:46.907920Z","signature_b64":"RRtErAbR9OeC24fgrlZ86T3lHf6p2P7mjevSp78SJGMjQFE2NzQ7iBtZBNNt1q01+mkE74BOpCX9EJppJnXEAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b292161ddf8c16bbfdb57c783ebf6de6b96b105467b76f481300ed245d84a95f","last_reissued_at":"2026-07-05T04:28:46.907414Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:28:46.907414Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adversarial Unlearning: Reducing Confidence Along Adversarial Directions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Amrith Setlur, Benjamin Eysenbach, Sergey Levine, Virginia Smith","submitted_at":"2022-06-03T02:26:24Z","abstract_excerpt":"Supervised learning methods trained with maximum likelihood objectives often overfit on training data. Most regularizers that prevent overfitting look to increase confidence on additional examples (e.g., data augmentation, adversarial training), or reduce it on training data (e.g., label smoothing). In this work we propose a complementary regularization strategy that reduces confidence on self-generated examples. The method, which we call RCAD (Reducing Confidence along Adversarial Directions), aims to reduce confidence on out-of-distribution examples lying along directions adversarially chose"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2206.01367","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/2206.01367/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":"2206.01367","created_at":"2026-07-05T04:28:46.907472+00:00"},{"alias_kind":"arxiv_version","alias_value":"2206.01367v1","created_at":"2026-07-05T04:28:46.907472+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2206.01367","created_at":"2026-07-05T04:28:46.907472+00:00"},{"alias_kind":"pith_short_12","alias_value":"WKJBMHO7RQLL","created_at":"2026-07-05T04:28:46.907472+00:00"},{"alias_kind":"pith_short_16","alias_value":"WKJBMHO7RQLLX7NV","created_at":"2026-07-05T04:28:46.907472+00:00"},{"alias_kind":"pith_short_8","alias_value":"WKJBMHO7","created_at":"2026-07-05T04:28:46.907472+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/WKJBMHO7RQLLX7NVPR4D5P3N42","json":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42.json","graph_json":"https://pith.science/api/pith-number/WKJBMHO7RQLLX7NVPR4D5P3N42/graph.json","events_json":"https://pith.science/api/pith-number/WKJBMHO7RQLLX7NVPR4D5P3N42/events.json","paper":"https://pith.science/paper/WKJBMHO7"},"agent_actions":{"view_html":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42","download_json":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42.json","view_paper":"https://pith.science/paper/WKJBMHO7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2206.01367&json=true","fetch_graph":"https://pith.science/api/pith-number/WKJBMHO7RQLLX7NVPR4D5P3N42/graph.json","fetch_events":"https://pith.science/api/pith-number/WKJBMHO7RQLLX7NVPR4D5P3N42/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42/action/storage_attestation","attest_author":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42/action/author_attestation","sign_citation":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42/action/citation_signature","submit_replication":"https://pith.science/pith/WKJBMHO7RQLLX7NVPR4D5P3N42/action/replication_record"}},"created_at":"2026-07-05T04:28:46.907472+00:00","updated_at":"2026-07-05T04:28:46.907472+00:00"}