{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:534EXLTKGKKWCIWYV7VT3YVRO7","short_pith_number":"pith:534EXLTK","schema_version":"1.0","canonical_sha256":"eef84bae6a32956122d8afeb3de2b177df8cce2fc1a1bdf4711038145f9faee4","source":{"kind":"arxiv","id":"2508.10243","version":1},"attestation_state":"computed","paper":{"title":"Pruning and Malicious Injection: A Retraining-Free Backdoor Attack on Transformer Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hao Wang, Mingxuan Sun, Taibiao Zhao, Xiangwei Zhou, Xiaobing Chen","submitted_at":"2025-08-14T00:13:22Z","abstract_excerpt":"Transformer models have demonstrated exceptional performance and have become indispensable in computer vision (CV) and natural language processing (NLP) tasks. However, recent studies reveal that transformers are susceptible to backdoor attacks. Prior backdoor attack methods typically rely on retraining with clean data or altering the model architecture, both of which can be resource-intensive and intrusive. In this paper, we propose Head-wise Pruning and Malicious Injection (HPMI), a novel retraining-free backdoor attack on transformers that does not alter the model's architecture. Our approa"},"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":"2508.10243","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-14T00:13:22Z","cross_cats_sorted":[],"title_canon_sha256":"7e9c911e49beeffb957bb87d3ab74a613f8e6ced7c3667ebe57c40c307673152","abstract_canon_sha256":"6f1852657a082974d04af6390d97866bb320a8d2a008f47429d7588f3b61ac60"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:53:51.976381Z","signature_b64":"bdeG0C7y6BCWZrz4hZe7h9JFaHfHdHUWPmLHKwUMPhOfyk4wF3rpPceWIG+5skE022cvnyxRLHXVJMXUGzFtDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eef84bae6a32956122d8afeb3de2b177df8cce2fc1a1bdf4711038145f9faee4","last_reissued_at":"2026-07-05T11:53:51.975695Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:53:51.975695Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pruning and Malicious Injection: A Retraining-Free Backdoor Attack on Transformer Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hao Wang, Mingxuan Sun, Taibiao Zhao, Xiangwei Zhou, Xiaobing Chen","submitted_at":"2025-08-14T00:13:22Z","abstract_excerpt":"Transformer models have demonstrated exceptional performance and have become indispensable in computer vision (CV) and natural language processing (NLP) tasks. However, recent studies reveal that transformers are susceptible to backdoor attacks. Prior backdoor attack methods typically rely on retraining with clean data or altering the model architecture, both of which can be resource-intensive and intrusive. In this paper, we propose Head-wise Pruning and Malicious Injection (HPMI), a novel retraining-free backdoor attack on transformers that does not alter the model's architecture. Our approa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.10243","kind":"arxiv","version":1},"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/2508.10243/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":"2508.10243","created_at":"2026-07-05T11:53:51.975763+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.10243v1","created_at":"2026-07-05T11:53:51.975763+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.10243","created_at":"2026-07-05T11:53:51.975763+00:00"},{"alias_kind":"pith_short_12","alias_value":"534EXLTKGKKW","created_at":"2026-07-05T11:53:51.975763+00:00"},{"alias_kind":"pith_short_16","alias_value":"534EXLTKGKKWCIWY","created_at":"2026-07-05T11:53:51.975763+00:00"},{"alias_kind":"pith_short_8","alias_value":"534EXLTK","created_at":"2026-07-05T11:53:51.975763+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.10242","citing_title":"Thermospheric Density, Composition, and Temperature from GOES-R/SUVI Solar Occultations","ref_index":1,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7","json":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7.json","graph_json":"https://pith.science/api/pith-number/534EXLTKGKKWCIWYV7VT3YVRO7/graph.json","events_json":"https://pith.science/api/pith-number/534EXLTKGKKWCIWYV7VT3YVRO7/events.json","paper":"https://pith.science/paper/534EXLTK"},"agent_actions":{"view_html":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7","download_json":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7.json","view_paper":"https://pith.science/paper/534EXLTK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.10243&json=true","fetch_graph":"https://pith.science/api/pith-number/534EXLTKGKKWCIWYV7VT3YVRO7/graph.json","fetch_events":"https://pith.science/api/pith-number/534EXLTKGKKWCIWYV7VT3YVRO7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7/action/storage_attestation","attest_author":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7/action/author_attestation","sign_citation":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7/action/citation_signature","submit_replication":"https://pith.science/pith/534EXLTKGKKWCIWYV7VT3YVRO7/action/replication_record"}},"created_at":"2026-07-05T11:53:51.975763+00:00","updated_at":"2026-07-05T11:53:51.975763+00:00"}