{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AZBKCJCASZJYUD2E3LAFOD57TD","short_pith_number":"pith:AZBKCJCA","schema_version":"1.0","canonical_sha256":"0642a1244096538a0f44dac0570fbf98cc185042d9ee0c218a6b291c31410833","source":{"kind":"arxiv","id":"2505.04468","version":2},"attestation_state":"computed","paper":{"title":"Fast Fourier Transform-Based Spectral and Temporal Gradient Filtering for Differential Privacy","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI","cs.IT","cs.NE","math.IT"],"primary_cat":"cs.LG","authors_text":"Hyeju Shin, Kyudan Jung, Seongwon Yun, Vincent-Daniel","submitted_at":"2025-05-07T14:38:58Z","abstract_excerpt":"Differential Privacy (DP) has emerged as a key framework for protecting sensitive data in machine learning, but standard DP-SGD often suffers from significant accuracy loss due to injected noise. To address this limitation, we introduce the FFT-Enhanced Kalman Filter (FFTKF), a differentially private optimization method that improves gradient quality while preserving $(\\varepsilon, \\delta)$-DP guarantees. FFTKF applies frequency-domain filtering to shift privacy noise into less informative high-frequency components, preserving the low-frequency gradient signals that carry most learning informa"},"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":"2505.04468","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-07T14:38:58Z","cross_cats_sorted":["cs.AI","cs.IT","cs.NE","math.IT"],"title_canon_sha256":"2f1e38104f49ab1669b688b74412cf00a17e965207080172f1efbf6d4b00fbb7","abstract_canon_sha256":"8026d625c00846cf85fa8783e4a344863486269f0d889ac08ff1ea8648e5e339"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:11:17.712197Z","signature_b64":"OJGbUdV66tZ62kA2eIrGnmfup+yUtYcZm2Tff+4T912nw2RcNDdd7BJb/OfXISD23S9jCNgQTbbklQi6n90rBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0642a1244096538a0f44dac0570fbf98cc185042d9ee0c218a6b291c31410833","last_reissued_at":"2026-07-05T12:11:17.711470Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:11:17.711470Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fast Fourier Transform-Based Spectral and Temporal Gradient Filtering for Differential Privacy","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI","cs.IT","cs.NE","math.IT"],"primary_cat":"cs.LG","authors_text":"Hyeju Shin, Kyudan Jung, Seongwon Yun, Vincent-Daniel","submitted_at":"2025-05-07T14:38:58Z","abstract_excerpt":"Differential Privacy (DP) has emerged as a key framework for protecting sensitive data in machine learning, but standard DP-SGD often suffers from significant accuracy loss due to injected noise. To address this limitation, we introduce the FFT-Enhanced Kalman Filter (FFTKF), a differentially private optimization method that improves gradient quality while preserving $(\\varepsilon, \\delta)$-DP guarantees. FFTKF applies frequency-domain filtering to shift privacy noise into less informative high-frequency components, preserving the low-frequency gradient signals that carry most learning informa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.04468","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/2505.04468/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":"2505.04468","created_at":"2026-07-05T12:11:17.711551+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.04468v2","created_at":"2026-07-05T12:11:17.711551+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.04468","created_at":"2026-07-05T12:11:17.711551+00:00"},{"alias_kind":"pith_short_12","alias_value":"AZBKCJCASZJY","created_at":"2026-07-05T12:11:17.711551+00:00"},{"alias_kind":"pith_short_16","alias_value":"AZBKCJCASZJYUD2E","created_at":"2026-07-05T12:11:17.711551+00:00"},{"alias_kind":"pith_short_8","alias_value":"AZBKCJCA","created_at":"2026-07-05T12:11:17.711551+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/AZBKCJCASZJYUD2E3LAFOD57TD","json":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD.json","graph_json":"https://pith.science/api/pith-number/AZBKCJCASZJYUD2E3LAFOD57TD/graph.json","events_json":"https://pith.science/api/pith-number/AZBKCJCASZJYUD2E3LAFOD57TD/events.json","paper":"https://pith.science/paper/AZBKCJCA"},"agent_actions":{"view_html":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD","download_json":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD.json","view_paper":"https://pith.science/paper/AZBKCJCA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.04468&json=true","fetch_graph":"https://pith.science/api/pith-number/AZBKCJCASZJYUD2E3LAFOD57TD/graph.json","fetch_events":"https://pith.science/api/pith-number/AZBKCJCASZJYUD2E3LAFOD57TD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD/action/storage_attestation","attest_author":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD/action/author_attestation","sign_citation":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD/action/citation_signature","submit_replication":"https://pith.science/pith/AZBKCJCASZJYUD2E3LAFOD57TD/action/replication_record"}},"created_at":"2026-07-05T12:11:17.711551+00:00","updated_at":"2026-07-05T12:11:17.711551+00:00"}