{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:CMJUZ4GTFKLRB3VBWIIACI5LWN","short_pith_number":"pith:CMJUZ4GT","schema_version":"1.0","canonical_sha256":"13134cf0d32a9710eea1b2100123abb35c3adc1128923f678f606ccc35caadb2","source":{"kind":"arxiv","id":"2005.12782","version":3},"attestation_state":"computed","paper":{"title":"A Protection against the Extraction of Neural Network Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Herv\\'e Chabanne, Linda Guiga, Vincent Despiegel","submitted_at":"2020-05-26T15:04:29Z","abstract_excerpt":"Given oracle access to a Neural Network (NN), it is possible to extract its underlying model. We here introduce a protection by adding parasitic layers which keep the underlying NN's predictions mostly unchanged while complexifying the task of reverse-engineering. Our countermeasure relies on approximating a noisy identity mapping with a Convolutional NN. We explain why the introduction of new parasitic layers complexifies the attacks. We report experiments regarding the performance and the accuracy of the protected NN."},"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":"2005.12782","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-05-26T15:04:29Z","cross_cats_sorted":["cs.CR","stat.ML"],"title_canon_sha256":"1f4a6606a4b735fbcc8ba923cd9ab81b67342b4f9d6ec5ff9d689d9dc2ea5c99","abstract_canon_sha256":"0300ca0358c90e3bf1a20c4c8b8260e797859313dbd2228379f6fbca41929b2d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:23:36.062477Z","signature_b64":"ueOF6klJ/tCL7fN/NsxVvR8aSr+Kpa4911Vs0P0wKTBhAJI9eeuYPLirSwbOgskkuC/daBsTG3VZ0iXvO/pGCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"13134cf0d32a9710eea1b2100123abb35c3adc1128923f678f606ccc35caadb2","last_reissued_at":"2026-07-05T01:23:36.061933Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:23:36.061933Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Protection against the Extraction of Neural Network Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Herv\\'e Chabanne, Linda Guiga, Vincent Despiegel","submitted_at":"2020-05-26T15:04:29Z","abstract_excerpt":"Given oracle access to a Neural Network (NN), it is possible to extract its underlying model. We here introduce a protection by adding parasitic layers which keep the underlying NN's predictions mostly unchanged while complexifying the task of reverse-engineering. Our countermeasure relies on approximating a noisy identity mapping with a Convolutional NN. We explain why the introduction of new parasitic layers complexifies the attacks. We report experiments regarding the performance and the accuracy of the protected NN."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2005.12782","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/2005.12782/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":"2005.12782","created_at":"2026-07-05T01:23:36.061988+00:00"},{"alias_kind":"arxiv_version","alias_value":"2005.12782v3","created_at":"2026-07-05T01:23:36.061988+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2005.12782","created_at":"2026-07-05T01:23:36.061988+00:00"},{"alias_kind":"pith_short_12","alias_value":"CMJUZ4GTFKLR","created_at":"2026-07-05T01:23:36.061988+00:00"},{"alias_kind":"pith_short_16","alias_value":"CMJUZ4GTFKLRB3VB","created_at":"2026-07-05T01:23:36.061988+00:00"},{"alias_kind":"pith_short_8","alias_value":"CMJUZ4GT","created_at":"2026-07-05T01:23:36.061988+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/CMJUZ4GTFKLRB3VBWIIACI5LWN","json":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN.json","graph_json":"https://pith.science/api/pith-number/CMJUZ4GTFKLRB3VBWIIACI5LWN/graph.json","events_json":"https://pith.science/api/pith-number/CMJUZ4GTFKLRB3VBWIIACI5LWN/events.json","paper":"https://pith.science/paper/CMJUZ4GT"},"agent_actions":{"view_html":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN","download_json":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN.json","view_paper":"https://pith.science/paper/CMJUZ4GT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2005.12782&json=true","fetch_graph":"https://pith.science/api/pith-number/CMJUZ4GTFKLRB3VBWIIACI5LWN/graph.json","fetch_events":"https://pith.science/api/pith-number/CMJUZ4GTFKLRB3VBWIIACI5LWN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN/action/storage_attestation","attest_author":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN/action/author_attestation","sign_citation":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN/action/citation_signature","submit_replication":"https://pith.science/pith/CMJUZ4GTFKLRB3VBWIIACI5LWN/action/replication_record"}},"created_at":"2026-07-05T01:23:36.061988+00:00","updated_at":"2026-07-05T01:23:36.061988+00:00"}