{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:XLHOHO2YOTKGIAKG6U4ED7E7GR","short_pith_number":"pith:XLHOHO2Y","schema_version":"1.0","canonical_sha256":"bacee3bb5874d4640146f53841fc9f34470bc633f67511b6e3f8146d7e4128e7","source":{"kind":"arxiv","id":"2411.09540","version":2},"attestation_state":"computed","paper":{"title":"Prompting the Unseen: Detecting Hidden Backdoors in Black-Box Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Chia-Mu Yu, Jia-Wei Chen, Zhi-Peng Zhang, Zi-Xuan Huang","submitted_at":"2024-11-14T15:56:11Z","abstract_excerpt":"Visual prompting (VP) is a new technique that adapts well-trained frozen models for source domain tasks to target domain tasks. This study examines VP's benefits for black-box model-level backdoor detection. The visual prompt in VP maps class subspaces between source and target domains. We identify a misalignment, termed class subspace inconsistency, between clean and poisoned datasets. Based on this, we introduce \\textsc{BProm}, a black-box model-level detection method to identify backdoors in suspicious models, if any. \\textsc{BProm} leverages the low classification accuracy of prompted mode"},"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":"2411.09540","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-11-14T15:56:11Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"b304177e622631a4ddcdb32efa7303b2a31bcbbb78a5e6d50fa835398ce3cded","abstract_canon_sha256":"8398decae5834ba4a893b331e6c156d0852f78d5eb16626a8632a731a33bae07"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:45:13.915170Z","signature_b64":"rAQj9d34/+K1A1yr6DZbSUf9TM8SwQfgsyrRZKhUlpqy9RIogS3Fs0ASYmeEEY9rvS1/gfc1D+iSrZbhWBnVDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bacee3bb5874d4640146f53841fc9f34470bc633f67511b6e3f8146d7e4128e7","last_reissued_at":"2026-07-05T10:45:13.914647Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:45:13.914647Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Prompting the Unseen: Detecting Hidden Backdoors in Black-Box Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Chia-Mu Yu, Jia-Wei Chen, Zhi-Peng Zhang, Zi-Xuan Huang","submitted_at":"2024-11-14T15:56:11Z","abstract_excerpt":"Visual prompting (VP) is a new technique that adapts well-trained frozen models for source domain tasks to target domain tasks. This study examines VP's benefits for black-box model-level backdoor detection. The visual prompt in VP maps class subspaces between source and target domains. We identify a misalignment, termed class subspace inconsistency, between clean and poisoned datasets. Based on this, we introduce \\textsc{BProm}, a black-box model-level detection method to identify backdoors in suspicious models, if any. \\textsc{BProm} leverages the low classification accuracy of prompted mode"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.09540","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/2411.09540/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":"2411.09540","created_at":"2026-07-05T10:45:13.914709+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.09540v2","created_at":"2026-07-05T10:45:13.914709+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.09540","created_at":"2026-07-05T10:45:13.914709+00:00"},{"alias_kind":"pith_short_12","alias_value":"XLHOHO2YOTKG","created_at":"2026-07-05T10:45:13.914709+00:00"},{"alias_kind":"pith_short_16","alias_value":"XLHOHO2YOTKGIAKG","created_at":"2026-07-05T10:45:13.914709+00:00"},{"alias_kind":"pith_short_8","alias_value":"XLHOHO2Y","created_at":"2026-07-05T10:45:13.914709+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/XLHOHO2YOTKGIAKG6U4ED7E7GR","json":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR.json","graph_json":"https://pith.science/api/pith-number/XLHOHO2YOTKGIAKG6U4ED7E7GR/graph.json","events_json":"https://pith.science/api/pith-number/XLHOHO2YOTKGIAKG6U4ED7E7GR/events.json","paper":"https://pith.science/paper/XLHOHO2Y"},"agent_actions":{"view_html":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR","download_json":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR.json","view_paper":"https://pith.science/paper/XLHOHO2Y","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.09540&json=true","fetch_graph":"https://pith.science/api/pith-number/XLHOHO2YOTKGIAKG6U4ED7E7GR/graph.json","fetch_events":"https://pith.science/api/pith-number/XLHOHO2YOTKGIAKG6U4ED7E7GR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR/action/storage_attestation","attest_author":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR/action/author_attestation","sign_citation":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR/action/citation_signature","submit_replication":"https://pith.science/pith/XLHOHO2YOTKGIAKG6U4ED7E7GR/action/replication_record"}},"created_at":"2026-07-05T10:45:13.914709+00:00","updated_at":"2026-07-05T10:45:13.914709+00:00"}