{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:N5K7X2FIKKS6DJ33DRLRJSBVDB","short_pith_number":"pith:N5K7X2FI","schema_version":"1.0","canonical_sha256":"6f55fbe8a852a5e1a77b1c5714c8351870e9cf858897f1b9af47a4b609ffe4ee","source":{"kind":"arxiv","id":"2506.02362","version":1},"attestation_state":"computed","paper":{"title":"MISLEADER: Defending against Model Extraction with Ensembles of Distilled Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Minxing Zheng, Shixiang Zhu, Xueqi Cheng, Yushun Dong","submitted_at":"2025-06-03T01:37:09Z","abstract_excerpt":"Model extraction attacks aim to replicate the functionality of a black-box model through query access, threatening the intellectual property (IP) of machine-learning-as-a-service (MLaaS) providers. Defending against such attacks is challenging, as it must balance efficiency, robustness, and utility preservation in the real-world scenario. Despite the recent advances, most existing defenses presume that attacker queries have out-of-distribution (OOD) samples, enabling them to detect and disrupt suspicious inputs. However, this assumption is increasingly unreliable, as modern models are trained "},"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":"2506.02362","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-06-03T01:37:09Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ab97fff2be878ddb8c85c644cb9fea536263c9e3d0b70b3c4b24127905bc5a0c","abstract_canon_sha256":"8f1de6dcf6f78e9575e9c938499e674f6070de67971ebe7650f04e9ef6e07926"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:14:37.468495Z","signature_b64":"VyU4Wc52gqPm/+4T17fp/AQBQdwv6wV6rhz94n9jc/xaCmYSvspg52qp34S+cLHWwTRYPQ61YNJq73EjR8d3BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6f55fbe8a852a5e1a77b1c5714c8351870e9cf858897f1b9af47a4b609ffe4ee","last_reissued_at":"2026-07-05T11:14:37.468017Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:14:37.468017Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MISLEADER: Defending against Model Extraction with Ensembles of Distilled Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Minxing Zheng, Shixiang Zhu, Xueqi Cheng, Yushun Dong","submitted_at":"2025-06-03T01:37:09Z","abstract_excerpt":"Model extraction attacks aim to replicate the functionality of a black-box model through query access, threatening the intellectual property (IP) of machine-learning-as-a-service (MLaaS) providers. Defending against such attacks is challenging, as it must balance efficiency, robustness, and utility preservation in the real-world scenario. Despite the recent advances, most existing defenses presume that attacker queries have out-of-distribution (OOD) samples, enabling them to detect and disrupt suspicious inputs. However, this assumption is increasingly unreliable, as modern models are trained "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.02362","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/2506.02362/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":"2506.02362","created_at":"2026-07-05T11:14:37.468079+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.02362v1","created_at":"2026-07-05T11:14:37.468079+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.02362","created_at":"2026-07-05T11:14:37.468079+00:00"},{"alias_kind":"pith_short_12","alias_value":"N5K7X2FIKKS6","created_at":"2026-07-05T11:14:37.468079+00:00"},{"alias_kind":"pith_short_16","alias_value":"N5K7X2FIKKS6DJ33","created_at":"2026-07-05T11:14:37.468079+00:00"},{"alias_kind":"pith_short_8","alias_value":"N5K7X2FI","created_at":"2026-07-05T11:14:37.468079+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.11301","citing_title":"LatentRouter: Can We Choose the Right Multimodal Model Before Seeing Its Answer?","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB","json":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB.json","graph_json":"https://pith.science/api/pith-number/N5K7X2FIKKS6DJ33DRLRJSBVDB/graph.json","events_json":"https://pith.science/api/pith-number/N5K7X2FIKKS6DJ33DRLRJSBVDB/events.json","paper":"https://pith.science/paper/N5K7X2FI"},"agent_actions":{"view_html":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB","download_json":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB.json","view_paper":"https://pith.science/paper/N5K7X2FI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.02362&json=true","fetch_graph":"https://pith.science/api/pith-number/N5K7X2FIKKS6DJ33DRLRJSBVDB/graph.json","fetch_events":"https://pith.science/api/pith-number/N5K7X2FIKKS6DJ33DRLRJSBVDB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB/action/storage_attestation","attest_author":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB/action/author_attestation","sign_citation":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB/action/citation_signature","submit_replication":"https://pith.science/pith/N5K7X2FIKKS6DJ33DRLRJSBVDB/action/replication_record"}},"created_at":"2026-07-05T11:14:37.468079+00:00","updated_at":"2026-07-05T11:14:37.468079+00:00"}