{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DNU2UI5FL3QGIG65OWTFCH73GJ","short_pith_number":"pith:DNU2UI5F","schema_version":"1.0","canonical_sha256":"1b69aa23a55ee0641bdd75a6511ffb324b4ad8ae648c170532651f51a2d25f9c","source":{"kind":"arxiv","id":"2605.21146","version":1},"attestation_state":"computed","paper":{"title":"Detecting Trojaned DNNs via Spectral Regression Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SE"],"primary_cat":"cs.CR","authors_text":"Jinhan Kim, Paolo Tonella, Samuele Pasini","submitted_at":"2026-05-20T13:19:27Z","abstract_excerpt":"Modern DNNs are repeatedly fine-tuned to incorporate new data and functionality. This evolutionary workflow introduces a security risk when updated data cannot be fully trusted, as adversaries may implant Trojans during fine-tuning. We present MIST, a Trojan detection approach that analyzes how a model's internal representations change during fine-tuning. Rather than attempting to reconstruct trigger conditions, MIST characterizes benign model evolution using pre-activation spectra and flags updates whose spectral deviations are inconsistent with this reference. This framing treats Trojan dete"},"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":"2605.21146","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-20T13:19:27Z","cross_cats_sorted":["cs.AI","cs.SE"],"title_canon_sha256":"c9ec6bf945288d356d9b75af585fd1afb8615c1f5a41b195d901255ccffd7820","abstract_canon_sha256":"07cef4ecff3e66ded6edb56fd958c62f98711959a51ad3e93c80c5fc91c12075"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:39.863577Z","signature_b64":"Ivp2jTKCoHaXP+s9KU8kpKXYXi1qRxLGWsoTGK+awu2gm56aB4duwPrmdn2uojFfWKHZftUOoXtz9HhOlN/kAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1b69aa23a55ee0641bdd75a6511ffb324b4ad8ae648c170532651f51a2d25f9c","last_reissued_at":"2026-05-21T01:05:39.862789Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:39.862789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Detecting Trojaned DNNs via Spectral Regression Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SE"],"primary_cat":"cs.CR","authors_text":"Jinhan Kim, Paolo Tonella, Samuele Pasini","submitted_at":"2026-05-20T13:19:27Z","abstract_excerpt":"Modern DNNs are repeatedly fine-tuned to incorporate new data and functionality. This evolutionary workflow introduces a security risk when updated data cannot be fully trusted, as adversaries may implant Trojans during fine-tuning. We present MIST, a Trojan detection approach that analyzes how a model's internal representations change during fine-tuning. Rather than attempting to reconstruct trigger conditions, MIST characterizes benign model evolution using pre-activation spectra and flags updates whose spectral deviations are inconsistent with this reference. This framing treats Trojan dete"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.21146","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/2605.21146/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":"2605.21146","created_at":"2026-05-21T01:05:39.862934+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.21146v1","created_at":"2026-05-21T01:05:39.862934+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.21146","created_at":"2026-05-21T01:05:39.862934+00:00"},{"alias_kind":"pith_short_12","alias_value":"DNU2UI5FL3QG","created_at":"2026-05-21T01:05:39.862934+00:00"},{"alias_kind":"pith_short_16","alias_value":"DNU2UI5FL3QGIG65","created_at":"2026-05-21T01:05:39.862934+00:00"},{"alias_kind":"pith_short_8","alias_value":"DNU2UI5F","created_at":"2026-05-21T01:05:39.862934+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/DNU2UI5FL3QGIG65OWTFCH73GJ","json":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ.json","graph_json":"https://pith.science/api/pith-number/DNU2UI5FL3QGIG65OWTFCH73GJ/graph.json","events_json":"https://pith.science/api/pith-number/DNU2UI5FL3QGIG65OWTFCH73GJ/events.json","paper":"https://pith.science/paper/DNU2UI5F"},"agent_actions":{"view_html":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ","download_json":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ.json","view_paper":"https://pith.science/paper/DNU2UI5F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.21146&json=true","fetch_graph":"https://pith.science/api/pith-number/DNU2UI5FL3QGIG65OWTFCH73GJ/graph.json","fetch_events":"https://pith.science/api/pith-number/DNU2UI5FL3QGIG65OWTFCH73GJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ/action/storage_attestation","attest_author":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ/action/author_attestation","sign_citation":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ/action/citation_signature","submit_replication":"https://pith.science/pith/DNU2UI5FL3QGIG65OWTFCH73GJ/action/replication_record"}},"created_at":"2026-05-21T01:05:39.862934+00:00","updated_at":"2026-05-21T01:05:39.862934+00:00"}