{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:YNAKXV2RWQ4XDMCPUDPHPV7GZP","short_pith_number":"pith:YNAKXV2R","schema_version":"1.0","canonical_sha256":"c340abd751b43971b04fa0de77d7e6cbf7b2c28834aba4e1dfd5aa5fc0bcaff3","source":{"kind":"arxiv","id":"2206.04723","version":1},"attestation_state":"computed","paper":{"title":"On the Unreasonable Effectiveness of Federated Averaging with Heterogeneous Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Gauri Joshi, Jianyu Wang, Rudrajit Das, Satyen Kale, Tong Zhang, Zheng Xu","submitted_at":"2022-06-09T18:25:25Z","abstract_excerpt":"Existing theory predicts that data heterogeneity will degrade the performance of the Federated Averaging (FedAvg) algorithm in federated learning. However, in practice, the simple FedAvg algorithm converges very well. This paper explains the seemingly unreasonable effectiveness of FedAvg that contradicts the previous theoretical predictions. We find that the key assumption of bounded gradient dissimilarity in previous theoretical analyses is too pessimistic to characterize data heterogeneity in practical applications. For a simple quadratic problem, we demonstrate there exist regimes where lar"},"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":"2206.04723","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-06-09T18:25:25Z","cross_cats_sorted":[],"title_canon_sha256":"38a53ced2ef399fd9a70d2650dedeff59c2c80d3356f8a31c588a58cda0ca82f","abstract_canon_sha256":"4a3ecc345d5df36139c22b8be50e8093c979292870eefd31a60b09ccff76d862"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:30:43.454527Z","signature_b64":"tGOIUBS3WZfZtH5/74AblW9BvYI/FOGR1Wyn3jRjy8UQVbfgLzfcskOjcrkQB/TN8Y/ALFAc6pQmSAkt96hJBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c340abd751b43971b04fa0de77d7e6cbf7b2c28834aba4e1dfd5aa5fc0bcaff3","last_reissued_at":"2026-07-05T04:30:43.454103Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:30:43.454103Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Unreasonable Effectiveness of Federated Averaging with Heterogeneous Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Gauri Joshi, Jianyu Wang, Rudrajit Das, Satyen Kale, Tong Zhang, Zheng Xu","submitted_at":"2022-06-09T18:25:25Z","abstract_excerpt":"Existing theory predicts that data heterogeneity will degrade the performance of the Federated Averaging (FedAvg) algorithm in federated learning. However, in practice, the simple FedAvg algorithm converges very well. This paper explains the seemingly unreasonable effectiveness of FedAvg that contradicts the previous theoretical predictions. We find that the key assumption of bounded gradient dissimilarity in previous theoretical analyses is too pessimistic to characterize data heterogeneity in practical applications. For a simple quadratic problem, we demonstrate there exist regimes where lar"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2206.04723","kind":"arxiv","version":1},"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/2206.04723/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":"2206.04723","created_at":"2026-07-05T04:30:43.454158+00:00"},{"alias_kind":"arxiv_version","alias_value":"2206.04723v1","created_at":"2026-07-05T04:30:43.454158+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2206.04723","created_at":"2026-07-05T04:30:43.454158+00:00"},{"alias_kind":"pith_short_12","alias_value":"YNAKXV2RWQ4X","created_at":"2026-07-05T04:30:43.454158+00:00"},{"alias_kind":"pith_short_16","alias_value":"YNAKXV2RWQ4XDMCP","created_at":"2026-07-05T04:30:43.454158+00:00"},{"alias_kind":"pith_short_8","alias_value":"YNAKXV2R","created_at":"2026-07-05T04:30:43.454158+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2409.03897","citing_title":"On the Convergence Rates of Federated Q-Learning across Heterogeneous Environments","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP","json":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP.json","graph_json":"https://pith.science/api/pith-number/YNAKXV2RWQ4XDMCPUDPHPV7GZP/graph.json","events_json":"https://pith.science/api/pith-number/YNAKXV2RWQ4XDMCPUDPHPV7GZP/events.json","paper":"https://pith.science/paper/YNAKXV2R"},"agent_actions":{"view_html":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP","download_json":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP.json","view_paper":"https://pith.science/paper/YNAKXV2R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2206.04723&json=true","fetch_graph":"https://pith.science/api/pith-number/YNAKXV2RWQ4XDMCPUDPHPV7GZP/graph.json","fetch_events":"https://pith.science/api/pith-number/YNAKXV2RWQ4XDMCPUDPHPV7GZP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP/action/storage_attestation","attest_author":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP/action/author_attestation","sign_citation":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP/action/citation_signature","submit_replication":"https://pith.science/pith/YNAKXV2RWQ4XDMCPUDPHPV7GZP/action/replication_record"}},"created_at":"2026-07-05T04:30:43.454158+00:00","updated_at":"2026-07-05T04:30:43.454158+00:00"}