{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VDKSECHBJKYLGSLZP6ZCI7YS7H","short_pith_number":"pith:VDKSECHB","schema_version":"1.0","canonical_sha256":"a8d52208e14ab0b349797fb2247f12f9eb3a0ee59d4994d3aeb2d15c8dbabc5a","source":{"kind":"arxiv","id":"2408.08868","version":3},"attestation_state":"computed","paper":{"title":"A Hassle-free Algorithm for Private Learning in Practice: Don't Use Tree Aggregation, Use BLTs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"H. Brendan McMahan, Yanxiang Zhang, Zheng Xu","submitted_at":"2024-08-16T17:52:22Z","abstract_excerpt":"The state-of-the-art for training on-device language models for mobile keyboard applications combines federated learning (FL) with differential privacy (DP) via the DP-Follow-the-Regularized-Leader (DP-FTRL) algorithm. Two variants of DP-FTRL are used in practice, tree aggregation and matrix factorization. However, tree aggregation suffers from significantly suboptimal privacy/utility tradeoffs, while matrix mechanisms require expensive optimization parameterized by hard-to-estimate-in-advance constants, and high runtime memory costs. This paper extends the recently introduced Buffered Linear "},"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":"2408.08868","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-08-16T17:52:22Z","cross_cats_sorted":[],"title_canon_sha256":"f8a88cd9247c69b033e6d65574fdfaeae334f4c79749a633e68e79801485f06a","abstract_canon_sha256":"8838ed14a410d00260072a299221cd9412af74a3a5cde47a34ab71b0135db70c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:12:19.357652Z","signature_b64":"RHOVyqSl/fdQonbUhFKUsc9pqBl5hwepXjC5+fst6yMnw5areJA/P9v5f5kQK7E8LLGYOpOz5Aenhi/SJpd0AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a8d52208e14ab0b349797fb2247f12f9eb3a0ee59d4994d3aeb2d15c8dbabc5a","last_reissued_at":"2026-07-05T11:12:19.357098Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:12:19.357098Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Hassle-free Algorithm for Private Learning in Practice: Don't Use Tree Aggregation, Use BLTs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"H. Brendan McMahan, Yanxiang Zhang, Zheng Xu","submitted_at":"2024-08-16T17:52:22Z","abstract_excerpt":"The state-of-the-art for training on-device language models for mobile keyboard applications combines federated learning (FL) with differential privacy (DP) via the DP-Follow-the-Regularized-Leader (DP-FTRL) algorithm. Two variants of DP-FTRL are used in practice, tree aggregation and matrix factorization. However, tree aggregation suffers from significantly suboptimal privacy/utility tradeoffs, while matrix mechanisms require expensive optimization parameterized by hard-to-estimate-in-advance constants, and high runtime memory costs. This paper extends the recently introduced Buffered Linear "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.08868","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/2408.08868/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":"2408.08868","created_at":"2026-07-05T11:12:19.357162+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.08868v3","created_at":"2026-07-05T11:12:19.357162+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.08868","created_at":"2026-07-05T11:12:19.357162+00:00"},{"alias_kind":"pith_short_12","alias_value":"VDKSECHBJKYL","created_at":"2026-07-05T11:12:19.357162+00:00"},{"alias_kind":"pith_short_16","alias_value":"VDKSECHBJKYLGSLZ","created_at":"2026-07-05T11:12:19.357162+00:00"},{"alias_kind":"pith_short_8","alias_value":"VDKSECHB","created_at":"2026-07-05T11:12:19.357162+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.20727","citing_title":"Edit-Neighboring Data Streams and Privacy under Continual Observation","ref_index":130,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H","json":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H.json","graph_json":"https://pith.science/api/pith-number/VDKSECHBJKYLGSLZP6ZCI7YS7H/graph.json","events_json":"https://pith.science/api/pith-number/VDKSECHBJKYLGSLZP6ZCI7YS7H/events.json","paper":"https://pith.science/paper/VDKSECHB"},"agent_actions":{"view_html":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H","download_json":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H.json","view_paper":"https://pith.science/paper/VDKSECHB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.08868&json=true","fetch_graph":"https://pith.science/api/pith-number/VDKSECHBJKYLGSLZP6ZCI7YS7H/graph.json","fetch_events":"https://pith.science/api/pith-number/VDKSECHBJKYLGSLZP6ZCI7YS7H/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H/action/storage_attestation","attest_author":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H/action/author_attestation","sign_citation":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H/action/citation_signature","submit_replication":"https://pith.science/pith/VDKSECHBJKYLGSLZP6ZCI7YS7H/action/replication_record"}},"created_at":"2026-07-05T11:12:19.357162+00:00","updated_at":"2026-07-05T11:12:19.357162+00:00"}