{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TBLOONOKQIX7L4CGZOPLODJBBL","short_pith_number":"pith:TBLOONOK","schema_version":"1.0","canonical_sha256":"9856e735ca822ff5f046cb9eb70d210afc21445e5a235c6b9970ab49ca3d413c","source":{"kind":"arxiv","id":"2506.01064","version":3},"attestation_state":"computed","paper":{"title":"Fighting Fire with Fire (F3): A Training-free and Efficient Visual Adversarial Example Purification Method in LVLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Di Wang, Jiansheng Chen, Ruobing Xie, Xingwu Sun, Yiqing Huang, Yudong Zhang, Yu Wang, Zhanhui Kang","submitted_at":"2025-06-01T16:07:30Z","abstract_excerpt":"Recent advances in large vision-language models (LVLMs) have showcased their remarkable capabilities across a wide range of multimodal vision-language tasks. However, these models remain vulnerable to visual adversarial attacks, which can substantially compromise their performance. In this paper, we introduce F3, a novel adversarial purification framework that employs a counterintuitive ``fighting fire with fire'' strategy: intentionally introducing simple perturbations to adversarial examples to mitigate their harmful effects. Specifically, F3 leverages cross-modal attentions derived from ran"},"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":"2506.01064","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-06-01T16:07:30Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"44a07cd54c3fd8e980874c79b04c0b9d84858e7425819bf4b6f2874bfefbe7c4","abstract_canon_sha256":"dd9a77b10e13d9a846dcadc470523405f2d91561d92cad07132b4be72c7bb7f7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:11:18.394992Z","signature_b64":"ndQgO0Vo3uQjso0ENvwb0PYmqp23XzoEIZDHi0/VJQnXiShmVcO7FZe8URQoMemVBfTfOo0gs4lNxE4oKX6GCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9856e735ca822ff5f046cb9eb70d210afc21445e5a235c6b9970ab49ca3d413c","last_reissued_at":"2026-07-05T12:11:18.394411Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:11:18.394411Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fighting Fire with Fire (F3): A Training-free and Efficient Visual Adversarial Example Purification Method in LVLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Di Wang, Jiansheng Chen, Ruobing Xie, Xingwu Sun, Yiqing Huang, Yudong Zhang, Yu Wang, Zhanhui Kang","submitted_at":"2025-06-01T16:07:30Z","abstract_excerpt":"Recent advances in large vision-language models (LVLMs) have showcased their remarkable capabilities across a wide range of multimodal vision-language tasks. However, these models remain vulnerable to visual adversarial attacks, which can substantially compromise their performance. In this paper, we introduce F3, a novel adversarial purification framework that employs a counterintuitive ``fighting fire with fire'' strategy: intentionally introducing simple perturbations to adversarial examples to mitigate their harmful effects. Specifically, F3 leverages cross-modal attentions derived from ran"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.01064","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/2506.01064/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":"2506.01064","created_at":"2026-07-05T12:11:18.394478+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.01064v3","created_at":"2026-07-05T12:11:18.394478+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.01064","created_at":"2026-07-05T12:11:18.394478+00:00"},{"alias_kind":"pith_short_12","alias_value":"TBLOONOKQIX7","created_at":"2026-07-05T12:11:18.394478+00:00"},{"alias_kind":"pith_short_16","alias_value":"TBLOONOKQIX7L4CG","created_at":"2026-07-05T12:11:18.394478+00:00"},{"alias_kind":"pith_short_8","alias_value":"TBLOONOK","created_at":"2026-07-05T12:11:18.394478+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/TBLOONOKQIX7L4CGZOPLODJBBL","json":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL.json","graph_json":"https://pith.science/api/pith-number/TBLOONOKQIX7L4CGZOPLODJBBL/graph.json","events_json":"https://pith.science/api/pith-number/TBLOONOKQIX7L4CGZOPLODJBBL/events.json","paper":"https://pith.science/paper/TBLOONOK"},"agent_actions":{"view_html":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL","download_json":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL.json","view_paper":"https://pith.science/paper/TBLOONOK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.01064&json=true","fetch_graph":"https://pith.science/api/pith-number/TBLOONOKQIX7L4CGZOPLODJBBL/graph.json","fetch_events":"https://pith.science/api/pith-number/TBLOONOKQIX7L4CGZOPLODJBBL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL/action/storage_attestation","attest_author":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL/action/author_attestation","sign_citation":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL/action/citation_signature","submit_replication":"https://pith.science/pith/TBLOONOKQIX7L4CGZOPLODJBBL/action/replication_record"}},"created_at":"2026-07-05T12:11:18.394478+00:00","updated_at":"2026-07-05T12:11:18.394478+00:00"}