{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XLBOVAPR7SL5VHD4Y5NIZC5UHF","short_pith_number":"pith:XLBOVAPR","schema_version":"1.0","canonical_sha256":"bac2ea81f1fc97da9c7cc75a8c8bb439711e268d0786640e4946a11a6a027a1b","source":{"kind":"arxiv","id":"2502.02410","version":3},"attestation_state":"computed","paper":{"title":"Privacy Amplification by Structured Subsampling for Deep Differentially Private Time Series Forecasting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Anderson Schneider, Jan Schuchardt, Jed Guzelkabaagac, Mina Dalirrooyfard, Stephan G\\\"unnemann, Yuriy Nevmyvaka","submitted_at":"2025-02-04T15:29:00Z","abstract_excerpt":"Many forms of sensitive data, such as web traffic, mobility data, or hospital occupancy, are inherently sequential. The standard method for training machine learning models while ensuring privacy for units of sensitive information, such as individual hospital visits, is differentially private stochastic gradient descent (DP-SGD). However, we observe in this work that the formal guarantees of DP-SGD are incompatible with time series specific tasks like forecasting, since they rely on the privacy amplification attained by training on small, unstructured batches sampled from an unstructured datas"},"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":"2502.02410","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-04T15:29:00Z","cross_cats_sorted":["cs.CR","stat.ML"],"title_canon_sha256":"c2502c3b737448f9cfccc0d22502ece554666a1d6ad737b841cab473740fceda","abstract_canon_sha256":"2ac3d5c112e84a461baf7695e6b23ddb80a1eae6b0927edf32fbdca9e1bcf259"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:48:00.415011Z","signature_b64":"liDv764dxvuPLhB9R+rHjeAgp/nkiIjnmBTeczZGoxCuy+ECSUNksu+jx5NEUFwDwHB3ZWKI/3G4EIUTmg3PBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bac2ea81f1fc97da9c7cc75a8c8bb439711e268d0786640e4946a11a6a027a1b","last_reissued_at":"2026-07-05T11:48:00.414468Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:48:00.414468Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Privacy Amplification by Structured Subsampling for Deep Differentially Private Time Series Forecasting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","stat.ML"],"primary_cat":"cs.LG","authors_text":"Anderson Schneider, Jan Schuchardt, Jed Guzelkabaagac, Mina Dalirrooyfard, Stephan G\\\"unnemann, Yuriy Nevmyvaka","submitted_at":"2025-02-04T15:29:00Z","abstract_excerpt":"Many forms of sensitive data, such as web traffic, mobility data, or hospital occupancy, are inherently sequential. The standard method for training machine learning models while ensuring privacy for units of sensitive information, such as individual hospital visits, is differentially private stochastic gradient descent (DP-SGD). However, we observe in this work that the formal guarantees of DP-SGD are incompatible with time series specific tasks like forecasting, since they rely on the privacy amplification attained by training on small, unstructured batches sampled from an unstructured datas"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.02410","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/2502.02410/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":"2502.02410","created_at":"2026-07-05T11:48:00.414533+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.02410v3","created_at":"2026-07-05T11:48:00.414533+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.02410","created_at":"2026-07-05T11:48:00.414533+00:00"},{"alias_kind":"pith_short_12","alias_value":"XLBOVAPR7SL5","created_at":"2026-07-05T11:48:00.414533+00:00"},{"alias_kind":"pith_short_16","alias_value":"XLBOVAPR7SL5VHD4","created_at":"2026-07-05T11:48:00.414533+00:00"},{"alias_kind":"pith_short_8","alias_value":"XLBOVAPR","created_at":"2026-07-05T11:48:00.414533+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.21780","citing_title":"Provable Robustness against Backdoor Attacks via the Primal-Dual Perspective on Differential Privacy","ref_index":63,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF","json":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF.json","graph_json":"https://pith.science/api/pith-number/XLBOVAPR7SL5VHD4Y5NIZC5UHF/graph.json","events_json":"https://pith.science/api/pith-number/XLBOVAPR7SL5VHD4Y5NIZC5UHF/events.json","paper":"https://pith.science/paper/XLBOVAPR"},"agent_actions":{"view_html":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF","download_json":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF.json","view_paper":"https://pith.science/paper/XLBOVAPR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.02410&json=true","fetch_graph":"https://pith.science/api/pith-number/XLBOVAPR7SL5VHD4Y5NIZC5UHF/graph.json","fetch_events":"https://pith.science/api/pith-number/XLBOVAPR7SL5VHD4Y5NIZC5UHF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF/action/storage_attestation","attest_author":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF/action/author_attestation","sign_citation":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF/action/citation_signature","submit_replication":"https://pith.science/pith/XLBOVAPR7SL5VHD4Y5NIZC5UHF/action/replication_record"}},"created_at":"2026-07-05T11:48:00.414533+00:00","updated_at":"2026-07-05T11:48:00.414533+00:00"}