{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3JNHJJOVXV4LRBRJGXUATCPFGZ","short_pith_number":"pith:3JNHJJOV","schema_version":"1.0","canonical_sha256":"da5a74a5d5bd78b8862935e80989e5366bdbe1ef1d6d7e4472778ecf3d1436e7","source":{"kind":"arxiv","id":"2504.04858","version":3},"attestation_state":"computed","paper":{"title":"Don't Lag, RAG: Training-Free Adversarial Detection Using RAG","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Moshe Sipper, Ofer Hadar, Raz Lapid, Roie Kazoom","submitted_at":"2025-04-07T09:14:47Z","abstract_excerpt":"Adversarial patch attacks pose a major threat to vision systems by embedding localized perturbations that mislead deep models. Traditional defense methods often require retraining or fine-tuning, making them impractical for real-world deployment. We propose a training-free Visual Retrieval-Augmented Generation (VRAG) framework that integrates Vision-Language Models (VLMs) for adversarial patch detection. By retrieving visually similar patches and images that resemble stored attacks in a continuously expanding database, VRAG performs generative reasoning to identify diverse attack types, all wi"},"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":"2504.04858","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-04-07T09:14:47Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"c6ebe62b026242c84dfdaa3859e54cf84e94b73140bb6b008a0d87a70c1b8ff4","abstract_canon_sha256":"313e0e3a02dacdb4c3ee370e739ea32e3abc76483fb66472c41e3aff301323ab"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:45:33.544654Z","signature_b64":"rBkQ1xi9gv8CnpAVfjaEy9Y96nHr1GwCBNimYpC31FVT4fzzF62vCTlNreEAPMgY0OUe9vpoPVOP9w7ghTqfDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"da5a74a5d5bd78b8862935e80989e5366bdbe1ef1d6d7e4472778ecf3d1436e7","last_reissued_at":"2026-07-05T11:45:33.544097Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:45:33.544097Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Don't Lag, RAG: Training-Free Adversarial Detection Using RAG","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Moshe Sipper, Ofer Hadar, Raz Lapid, Roie Kazoom","submitted_at":"2025-04-07T09:14:47Z","abstract_excerpt":"Adversarial patch attacks pose a major threat to vision systems by embedding localized perturbations that mislead deep models. Traditional defense methods often require retraining or fine-tuning, making them impractical for real-world deployment. We propose a training-free Visual Retrieval-Augmented Generation (VRAG) framework that integrates Vision-Language Models (VLMs) for adversarial patch detection. By retrieving visually similar patches and images that resemble stored attacks in a continuously expanding database, VRAG performs generative reasoning to identify diverse attack types, all wi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.04858","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/2504.04858/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":"2504.04858","created_at":"2026-07-05T11:45:33.544164+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.04858v3","created_at":"2026-07-05T11:45:33.544164+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.04858","created_at":"2026-07-05T11:45:33.544164+00:00"},{"alias_kind":"pith_short_12","alias_value":"3JNHJJOVXV4L","created_at":"2026-07-05T11:45:33.544164+00:00"},{"alias_kind":"pith_short_16","alias_value":"3JNHJJOVXV4LRBRJ","created_at":"2026-07-05T11:45:33.544164+00:00"},{"alias_kind":"pith_short_8","alias_value":"3JNHJJOV","created_at":"2026-07-05T11:45:33.544164+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.20623","citing_title":"RSRCC: A Remote Sensing Regional Change Comprehension Benchmark Constructed via Retrieval-Augmented Best-of-N Ranking","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ","json":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ.json","graph_json":"https://pith.science/api/pith-number/3JNHJJOVXV4LRBRJGXUATCPFGZ/graph.json","events_json":"https://pith.science/api/pith-number/3JNHJJOVXV4LRBRJGXUATCPFGZ/events.json","paper":"https://pith.science/paper/3JNHJJOV"},"agent_actions":{"view_html":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ","download_json":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ.json","view_paper":"https://pith.science/paper/3JNHJJOV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.04858&json=true","fetch_graph":"https://pith.science/api/pith-number/3JNHJJOVXV4LRBRJGXUATCPFGZ/graph.json","fetch_events":"https://pith.science/api/pith-number/3JNHJJOVXV4LRBRJGXUATCPFGZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ/action/storage_attestation","attest_author":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ/action/author_attestation","sign_citation":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ/action/citation_signature","submit_replication":"https://pith.science/pith/3JNHJJOVXV4LRBRJGXUATCPFGZ/action/replication_record"}},"created_at":"2026-07-05T11:45:33.544164+00:00","updated_at":"2026-07-05T11:45:33.544164+00:00"}