{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:3FFDZBZKPSWZJKNVPPSNZSSMT6","short_pith_number":"pith:3FFDZBZK","schema_version":"1.0","canonical_sha256":"d94a3c872a7cad94a9b57be4dcca4c9f898694c4e1c494282c855f9ed3236cfb","source":{"kind":"arxiv","id":"2405.15234","version":3},"attestation_state":"computed","paper":{"title":"Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CV","authors_text":"Chongyu Fan, Jiancheng Liu, Jinghan Jia, Ke Ding, Mingyi Hong, Sijia Liu, Xin Chen, Yihua Zhang, Yimeng Zhang","submitted_at":"2024-05-24T05:47:23Z","abstract_excerpt":"Diffusion models (DMs) have achieved remarkable success in text-to-image generation, but they also pose safety risks, such as the potential generation of harmful content and copyright violations. The techniques of machine unlearning, also known as concept erasing, have been developed to address these risks. However, these techniques remain vulnerable to adversarial prompt attacks, which can prompt DMs post-unlearning to regenerate undesired images containing concepts (such as nudity) meant to be erased. This work aims to enhance the robustness of concept erasing by integrating the principle of"},"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":"2405.15234","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-24T05:47:23Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"696ffb03a44e3d4c6eaf080e42abfba9326f672a3330121b2b4badce7685d23c","abstract_canon_sha256":"d251a51325ba41ac1d6bf586f848b7276a368ef91cc18ae5c85af550e52f4f61"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:17:42.294334Z","signature_b64":"+t7JwAyTIGTa8JMcebI1BIjzSGHmTB3PUWox7dnwFbDYSEvYCKI74DGTdbV7KUsD8AMImAninEyhKR6mG5flBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d94a3c872a7cad94a9b57be4dcca4c9f898694c4e1c494282c855f9ed3236cfb","last_reissued_at":"2026-07-05T09:17:42.293845Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:17:42.293845Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CV","authors_text":"Chongyu Fan, Jiancheng Liu, Jinghan Jia, Ke Ding, Mingyi Hong, Sijia Liu, Xin Chen, Yihua Zhang, Yimeng Zhang","submitted_at":"2024-05-24T05:47:23Z","abstract_excerpt":"Diffusion models (DMs) have achieved remarkable success in text-to-image generation, but they also pose safety risks, such as the potential generation of harmful content and copyright violations. The techniques of machine unlearning, also known as concept erasing, have been developed to address these risks. However, these techniques remain vulnerable to adversarial prompt attacks, which can prompt DMs post-unlearning to regenerate undesired images containing concepts (such as nudity) meant to be erased. This work aims to enhance the robustness of concept erasing by integrating the principle of"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.15234","kind":"arxiv","version":3},"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/2405.15234/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":"2405.15234","created_at":"2026-07-05T09:17:42.293903+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.15234v3","created_at":"2026-07-05T09:17:42.293903+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.15234","created_at":"2026-07-05T09:17:42.293903+00:00"},{"alias_kind":"pith_short_12","alias_value":"3FFDZBZKPSWZ","created_at":"2026-07-05T09:17:42.293903+00:00"},{"alias_kind":"pith_short_16","alias_value":"3FFDZBZKPSWZJKNV","created_at":"2026-07-05T09:17:42.293903+00:00"},{"alias_kind":"pith_short_8","alias_value":"3FFDZBZK","created_at":"2026-07-05T09:17:42.293903+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07907","citing_title":"Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2606.25548","citing_title":"Concept Removal for Frontier Image Generative Models","ref_index":127,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02453","citing_title":"Initialization is Half the Battle: Generating Diverse Images from a Guidance Potential Posterior","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2501.03544","citing_title":"PromptGuard: Soft Prompt-Guided Unsafe Content Moderation for Text-to-Image Models","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.25119","citing_title":"Evaluation without Generation: Non-Generative Assessment of Harmful Model Specialization with Applications to CSAM","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17396","citing_title":"Representation-Guided Parameter-Efficient LLM Unlearning","ref_index":198,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6","json":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6.json","graph_json":"https://pith.science/api/pith-number/3FFDZBZKPSWZJKNVPPSNZSSMT6/graph.json","events_json":"https://pith.science/api/pith-number/3FFDZBZKPSWZJKNVPPSNZSSMT6/events.json","paper":"https://pith.science/paper/3FFDZBZK"},"agent_actions":{"view_html":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6","download_json":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6.json","view_paper":"https://pith.science/paper/3FFDZBZK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.15234&json=true","fetch_graph":"https://pith.science/api/pith-number/3FFDZBZKPSWZJKNVPPSNZSSMT6/graph.json","fetch_events":"https://pith.science/api/pith-number/3FFDZBZKPSWZJKNVPPSNZSSMT6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6/action/storage_attestation","attest_author":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6/action/author_attestation","sign_citation":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6/action/citation_signature","submit_replication":"https://pith.science/pith/3FFDZBZKPSWZJKNVPPSNZSSMT6/action/replication_record"}},"created_at":"2026-07-05T09:17:42.293903+00:00","updated_at":"2026-07-05T09:17:42.293903+00:00"}