{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:H2EFE46NKR2YD5BFDL63SUHTVT","short_pith_number":"pith:H2EFE46N","schema_version":"1.0","canonical_sha256":"3e885273cd547581f4251afdb950f3acec99d05668f469f29571d1b77ff4e24c","source":{"kind":"arxiv","id":"2210.01864","version":2},"attestation_state":"computed","paper":{"title":"Recycling Scraps: Improving Private Learning by Leveraging Intermediate Checkpoints","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Abhradeep Thakurta, Arun Ganesh, Om Thakkar, Rajiv Mathews, Shuang Song, Virat Shejwalkar, Xinyi Zheng, Yarong Mu","submitted_at":"2022-10-04T19:21:00Z","abstract_excerpt":"In this work, we focus on improving the accuracy-variance trade-off for state-of-the-art differentially private machine learning (DP ML) methods. First, we design a general framework that uses aggregates of intermediate checkpoints \\emph{during training} to increase the accuracy of DP ML techniques. Specifically, we demonstrate that training over aggregates can provide significant gains in prediction accuracy over the existing state-of-the-art for StackOverflow, CIFAR10 and CIFAR100 datasets. For instance, we improve the state-of-the-art DP StackOverflow accuracies to 22.74\\% (+2.06\\% relative"},"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":"2210.01864","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2022-10-04T19:21:00Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"705c9770216df0a0dde228d7a3c1b5b6a937d963d202657a8e1dd420967fde29","abstract_canon_sha256":"464a9159f2dd9f00909a8efd5b05807100a2ba2a45e234a3067a969a60381846"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:07:42.713950Z","signature_b64":"5rlqrQ5rNPs/y3WTJIGXWFAN4Qfp7cPC8VjWbHv3N4PsDTBMoNKl/rMDk6xJl/iMWbZ80yWD2mNt6AwdzvI3Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3e885273cd547581f4251afdb950f3acec99d05668f469f29571d1b77ff4e24c","last_reissued_at":"2026-07-05T09:07:42.713457Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:07:42.713457Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Recycling Scraps: Improving Private Learning by Leveraging Intermediate Checkpoints","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Abhradeep Thakurta, Arun Ganesh, Om Thakkar, Rajiv Mathews, Shuang Song, Virat Shejwalkar, Xinyi Zheng, Yarong Mu","submitted_at":"2022-10-04T19:21:00Z","abstract_excerpt":"In this work, we focus on improving the accuracy-variance trade-off for state-of-the-art differentially private machine learning (DP ML) methods. First, we design a general framework that uses aggregates of intermediate checkpoints \\emph{during training} to increase the accuracy of DP ML techniques. Specifically, we demonstrate that training over aggregates can provide significant gains in prediction accuracy over the existing state-of-the-art for StackOverflow, CIFAR10 and CIFAR100 datasets. For instance, we improve the state-of-the-art DP StackOverflow accuracies to 22.74\\% (+2.06\\% relative"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.01864","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/2210.01864/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":"2210.01864","created_at":"2026-07-05T09:07:42.713516+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.01864v2","created_at":"2026-07-05T09:07:42.713516+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.01864","created_at":"2026-07-05T09:07:42.713516+00:00"},{"alias_kind":"pith_short_12","alias_value":"H2EFE46NKR2Y","created_at":"2026-07-05T09:07:42.713516+00:00"},{"alias_kind":"pith_short_16","alias_value":"H2EFE46NKR2YD5BF","created_at":"2026-07-05T09:07:42.713516+00:00"},{"alias_kind":"pith_short_8","alias_value":"H2EFE46N","created_at":"2026-07-05T09:07:42.713516+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.20985","citing_title":"Differentially Private Model Merging","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20985","citing_title":"Differentially Private Model Merging","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT","json":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT.json","graph_json":"https://pith.science/api/pith-number/H2EFE46NKR2YD5BFDL63SUHTVT/graph.json","events_json":"https://pith.science/api/pith-number/H2EFE46NKR2YD5BFDL63SUHTVT/events.json","paper":"https://pith.science/paper/H2EFE46N"},"agent_actions":{"view_html":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT","download_json":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT.json","view_paper":"https://pith.science/paper/H2EFE46N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.01864&json=true","fetch_graph":"https://pith.science/api/pith-number/H2EFE46NKR2YD5BFDL63SUHTVT/graph.json","fetch_events":"https://pith.science/api/pith-number/H2EFE46NKR2YD5BFDL63SUHTVT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT/action/storage_attestation","attest_author":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT/action/author_attestation","sign_citation":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT/action/citation_signature","submit_replication":"https://pith.science/pith/H2EFE46NKR2YD5BFDL63SUHTVT/action/replication_record"}},"created_at":"2026-07-05T09:07:42.713516+00:00","updated_at":"2026-07-05T09:07:42.713516+00:00"}