{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:KQFRVABRBMODFIYOIILWAEI3WF","short_pith_number":"pith:KQFRVABR","schema_version":"1.0","canonical_sha256":"540b1a80310b1c32a30e421760111bb17ad9bc32a9a8ead30d6bb5d31f13fdb5","source":{"kind":"arxiv","id":"2308.12439","version":2},"attestation_state":"computed","paper":{"title":"BaDExpert: Extracting Backdoor Functionality for Accurate Backdoor Input Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CR","authors_text":"Jiachen T. Wang, Ping He, Prateek Mittal, Tinghao Xie, Xiangyu Qi, Yiming Li","submitted_at":"2023-08-23T21:47:06Z","abstract_excerpt":"We present a novel defense, against backdoor attacks on Deep Neural Networks (DNNs), wherein adversaries covertly implant malicious behaviors (backdoors) into DNNs. Our defense falls within the category of post-development defenses that operate independently of how the model was generated. The proposed defense is built upon a novel reverse engineering approach that can directly extract backdoor functionality of a given backdoored model to a backdoor expert model. The approach is straightforward -- finetuning the backdoored model over a small set of intentionally mislabeled clean samples, such "},"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":"2308.12439","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-08-23T21:47:06Z","cross_cats_sorted":["cs.AI","cs.CV","cs.LG"],"title_canon_sha256":"acc615d60acddf79dc389d3a952218ca26ccb73029a8d1039c44dc75363ab92a","abstract_canon_sha256":"451cbb969724e70d7dd43eef73b2d705a800353598374ff0c6032c45ffe5c949"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:57:25.485653Z","signature_b64":"HcOBVAFuT21ZMerKjnxojBnxksxO2tsw8XmQt35ICuU4OOMssL9FIC1VDlRDhkKWV5CMDEa5mc7w51FWVUyUCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"540b1a80310b1c32a30e421760111bb17ad9bc32a9a8ead30d6bb5d31f13fdb5","last_reissued_at":"2026-07-05T06:57:25.485156Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:57:25.485156Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BaDExpert: Extracting Backdoor Functionality for Accurate Backdoor Input Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CR","authors_text":"Jiachen T. Wang, Ping He, Prateek Mittal, Tinghao Xie, Xiangyu Qi, Yiming Li","submitted_at":"2023-08-23T21:47:06Z","abstract_excerpt":"We present a novel defense, against backdoor attacks on Deep Neural Networks (DNNs), wherein adversaries covertly implant malicious behaviors (backdoors) into DNNs. Our defense falls within the category of post-development defenses that operate independently of how the model was generated. The proposed defense is built upon a novel reverse engineering approach that can directly extract backdoor functionality of a given backdoored model to a backdoor expert model. The approach is straightforward -- finetuning the backdoored model over a small set of intentionally mislabeled clean samples, such "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.12439","kind":"arxiv","version":2},"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/2308.12439/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":"2308.12439","created_at":"2026-07-05T06:57:25.485220+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.12439v2","created_at":"2026-07-05T06:57:25.485220+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.12439","created_at":"2026-07-05T06:57:25.485220+00:00"},{"alias_kind":"pith_short_12","alias_value":"KQFRVABRBMOD","created_at":"2026-07-05T06:57:25.485220+00:00"},{"alias_kind":"pith_short_16","alias_value":"KQFRVABRBMODFIYO","created_at":"2026-07-05T06:57:25.485220+00:00"},{"alias_kind":"pith_short_8","alias_value":"KQFRVABR","created_at":"2026-07-05T06:57:25.485220+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11648","citing_title":"Dummy Backdoor as a Defense: Removing Unknown Backdoors via Shared Internal Mechanisms for Generative LLMs","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21146","citing_title":"Detecting Trojaned DNNs via Spectral Regression Analysis","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF","json":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF.json","graph_json":"https://pith.science/api/pith-number/KQFRVABRBMODFIYOIILWAEI3WF/graph.json","events_json":"https://pith.science/api/pith-number/KQFRVABRBMODFIYOIILWAEI3WF/events.json","paper":"https://pith.science/paper/KQFRVABR"},"agent_actions":{"view_html":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF","download_json":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF.json","view_paper":"https://pith.science/paper/KQFRVABR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.12439&json=true","fetch_graph":"https://pith.science/api/pith-number/KQFRVABRBMODFIYOIILWAEI3WF/graph.json","fetch_events":"https://pith.science/api/pith-number/KQFRVABRBMODFIYOIILWAEI3WF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF/action/storage_attestation","attest_author":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF/action/author_attestation","sign_citation":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF/action/citation_signature","submit_replication":"https://pith.science/pith/KQFRVABRBMODFIYOIILWAEI3WF/action/replication_record"}},"created_at":"2026-07-05T06:57:25.485220+00:00","updated_at":"2026-07-05T06:57:25.485220+00:00"}