{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:4ZGLZLWDXQBZGC2OBQCE5QPLBX","short_pith_number":"pith:4ZGLZLWD","schema_version":"1.0","canonical_sha256":"e64cbcaec3bc03930b4e0c044ec1eb0dcc75b4cec171eaad2280a1415528af7f","source":{"kind":"arxiv","id":"2107.12342","version":2},"attestation_state":"computed","paper":{"title":"Sisyphus: A Cautionary Tale of Using Low-Degree Polynomial Activations in Privacy-Preserving Deep Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Brandon Reagen, Karthik Garimella, Nandan Kumar Jha","submitted_at":"2021-07-26T17:33:56Z","abstract_excerpt":"Privacy concerns in client-server machine learning have given rise to private inference (PI), where neural inference occurs directly on encrypted inputs. PI protects clients' personal data and the server's intellectual property. A common practice in PI is to use garbled circuits to compute nonlinear functions privately, namely ReLUs. However, garbled circuits suffer from high storage, bandwidth, and latency costs. To mitigate these issues, PI-friendly polynomial activation functions have been employed to replace ReLU. In this work, we ask: Is it feasible to substitute all ReLUs with low-degree"},"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":"2107.12342","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-07-26T17:33:56Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"5b27dce75a9e3c0bff5ac4c08dac8c95bcc9b6956e5b14427ceb5156be7b7953","abstract_canon_sha256":"1e744ab87213666d3eabced090f22d9097db2303174e33595968f46e46e11bb5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:28:40.439422Z","signature_b64":"FfLzRJDGP/kYVttvFgxSdeAaUCAfKQFfF7vXwRc7JEopINpKL4laH8nwfw+yWNxrI27pXKlm8ZZfGa0odN24Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e64cbcaec3bc03930b4e0c044ec1eb0dcc75b4cec171eaad2280a1415528af7f","last_reissued_at":"2026-07-05T03:28:40.439008Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:28:40.439008Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sisyphus: A Cautionary Tale of Using Low-Degree Polynomial Activations in Privacy-Preserving Deep Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Brandon Reagen, Karthik Garimella, Nandan Kumar Jha","submitted_at":"2021-07-26T17:33:56Z","abstract_excerpt":"Privacy concerns in client-server machine learning have given rise to private inference (PI), where neural inference occurs directly on encrypted inputs. PI protects clients' personal data and the server's intellectual property. A common practice in PI is to use garbled circuits to compute nonlinear functions privately, namely ReLUs. However, garbled circuits suffer from high storage, bandwidth, and latency costs. To mitigate these issues, PI-friendly polynomial activation functions have been employed to replace ReLU. In this work, we ask: Is it feasible to substitute all ReLUs with low-degree"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.12342","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/2107.12342/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":"2107.12342","created_at":"2026-07-05T03:28:40.439058+00:00"},{"alias_kind":"arxiv_version","alias_value":"2107.12342v2","created_at":"2026-07-05T03:28:40.439058+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.12342","created_at":"2026-07-05T03:28:40.439058+00:00"},{"alias_kind":"pith_short_12","alias_value":"4ZGLZLWDXQBZ","created_at":"2026-07-05T03:28:40.439058+00:00"},{"alias_kind":"pith_short_16","alias_value":"4ZGLZLWDXQBZGC2O","created_at":"2026-07-05T03:28:40.439058+00:00"},{"alias_kind":"pith_short_8","alias_value":"4ZGLZLWD","created_at":"2026-07-05T03:28:40.439058+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.09607","citing_title":"Efficient and High-Accuracy Private CNN Inference with Helper-Assisted Malicious Security","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX","json":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX.json","graph_json":"https://pith.science/api/pith-number/4ZGLZLWDXQBZGC2OBQCE5QPLBX/graph.json","events_json":"https://pith.science/api/pith-number/4ZGLZLWDXQBZGC2OBQCE5QPLBX/events.json","paper":"https://pith.science/paper/4ZGLZLWD"},"agent_actions":{"view_html":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX","download_json":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX.json","view_paper":"https://pith.science/paper/4ZGLZLWD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2107.12342&json=true","fetch_graph":"https://pith.science/api/pith-number/4ZGLZLWDXQBZGC2OBQCE5QPLBX/graph.json","fetch_events":"https://pith.science/api/pith-number/4ZGLZLWDXQBZGC2OBQCE5QPLBX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX/action/storage_attestation","attest_author":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX/action/author_attestation","sign_citation":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX/action/citation_signature","submit_replication":"https://pith.science/pith/4ZGLZLWDXQBZGC2OBQCE5QPLBX/action/replication_record"}},"created_at":"2026-07-05T03:28:40.439058+00:00","updated_at":"2026-07-05T03:28:40.439058+00:00"}