{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:WRMY7ALCWSZSFGUN62EN7DF4VM","short_pith_number":"pith:WRMY7ALC","schema_version":"1.0","canonical_sha256":"b4598f8162b4b3229a8df688df8cbcab2e19d5b996c1ede860c1238d14800704","source":{"kind":"arxiv","id":"2304.14836","version":2},"attestation_state":"computed","paper":{"title":"Training Large Scale Polynomial CNNs for E2E Inference over Homomorphic Encryption","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR"],"primary_cat":"cs.LG","authors_text":"Eyal Kushnir, Gilad Ezov, Itamar Zimerman, Jenny Lerner, Moran Baruch, Nir Drucker, Omri Soceanu, Yoav Goldberg","submitted_at":"2023-04-26T20:41:37Z","abstract_excerpt":"Training large-scale CNNs that during inference can be run under Homomorphic Encryption (HE) is challenging due to the need to use only polynomial operations. This limits HE-based solutions adoption. We address this challenge and pioneer in providing a novel training method for large polynomial CNNs such as ResNet-152 and ConvNeXt models, and achieve promising accuracy on encrypted samples on large-scale dataset such as ImageNet. Additionally, we provide optimization insights regarding activation functions and skip-connection latency impacts, enhancing HE-based evaluation efficiency. Finally, "},"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":"2304.14836","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-04-26T20:41:37Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"39cc796f329bb009546772f22048b5e1dca23949cc06cd59188b8c7de5bf23e5","abstract_canon_sha256":"7fa8528c8b0cbece26899ebc516b72e42abe3085086d983d64808ae3ba435eda"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:19:19.504573Z","signature_b64":"fSC6K3dqLW3OJ++U9buNQ8S05oPY2i/JrNTNUvE5OojQPXPZbWyeSGKSCq1EUW3rbzzKVgu+h7Ad+djQwFBTCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b4598f8162b4b3229a8df688df8cbcab2e19d5b996c1ede860c1238d14800704","last_reissued_at":"2026-07-05T06:19:19.504139Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:19:19.504139Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Training Large Scale Polynomial CNNs for E2E Inference over Homomorphic Encryption","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR"],"primary_cat":"cs.LG","authors_text":"Eyal Kushnir, Gilad Ezov, Itamar Zimerman, Jenny Lerner, Moran Baruch, Nir Drucker, Omri Soceanu, Yoav Goldberg","submitted_at":"2023-04-26T20:41:37Z","abstract_excerpt":"Training large-scale CNNs that during inference can be run under Homomorphic Encryption (HE) is challenging due to the need to use only polynomial operations. This limits HE-based solutions adoption. We address this challenge and pioneer in providing a novel training method for large polynomial CNNs such as ResNet-152 and ConvNeXt models, and achieve promising accuracy on encrypted samples on large-scale dataset such as ImageNet. Additionally, we provide optimization insights regarding activation functions and skip-connection latency impacts, enhancing HE-based evaluation efficiency. Finally, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.14836","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/2304.14836/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":"2304.14836","created_at":"2026-07-05T06:19:19.504199+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.14836v2","created_at":"2026-07-05T06:19:19.504199+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.14836","created_at":"2026-07-05T06:19:19.504199+00:00"},{"alias_kind":"pith_short_12","alias_value":"WRMY7ALCWSZS","created_at":"2026-07-05T06:19:19.504199+00:00"},{"alias_kind":"pith_short_16","alias_value":"WRMY7ALCWSZSFGUN","created_at":"2026-07-05T06:19:19.504199+00:00"},{"alias_kind":"pith_short_8","alias_value":"WRMY7ALC","created_at":"2026-07-05T06:19:19.504199+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2410.09457","citing_title":"Power-Softmax: Towards Secure LLM Inference over Encrypted Data","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM","json":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM.json","graph_json":"https://pith.science/api/pith-number/WRMY7ALCWSZSFGUN62EN7DF4VM/graph.json","events_json":"https://pith.science/api/pith-number/WRMY7ALCWSZSFGUN62EN7DF4VM/events.json","paper":"https://pith.science/paper/WRMY7ALC"},"agent_actions":{"view_html":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM","download_json":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM.json","view_paper":"https://pith.science/paper/WRMY7ALC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.14836&json=true","fetch_graph":"https://pith.science/api/pith-number/WRMY7ALCWSZSFGUN62EN7DF4VM/graph.json","fetch_events":"https://pith.science/api/pith-number/WRMY7ALCWSZSFGUN62EN7DF4VM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM/action/storage_attestation","attest_author":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM/action/author_attestation","sign_citation":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM/action/citation_signature","submit_replication":"https://pith.science/pith/WRMY7ALCWSZSFGUN62EN7DF4VM/action/replication_record"}},"created_at":"2026-07-05T06:19:19.504199+00:00","updated_at":"2026-07-05T06:19:19.504199+00:00"}