{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:VZAZBDIC752UTMEZASMQWQ7JBA","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"ea16202aaf1c40a1a276d37b9147b273854430fed81d2d85e909613bdca48229","cross_cats_sorted":["cs.CV","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-09-13T17:26:20Z","title_canon_sha256":"bdf8dc665285f408ac522b864cb845415d7e8dfc30ccbca9450d38eaf2efea69"},"schema_version":"1.0","source":{"id":"1909.06335","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1909.06335","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"arxiv_version","alias_value":"1909.06335v1","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1909.06335","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"pith_short_12","alias_value":"VZAZBDIC752U","created_at":"2026-05-18T12:33:30Z"},{"alias_kind":"pith_short_16","alias_value":"VZAZBDIC752UTMEZ","created_at":"2026-05-18T12:33:30Z"},{"alias_kind":"pith_short_8","alias_value":"VZAZBDIC","created_at":"2026-05-18T12:33:30Z"}],"graph_snapshots":[{"event_id":"sha256:ecf46b04836f560b3a6c3bdc296e5560b763a5088bf77027347c6a269ea6a807","target":"graph","created_at":"2026-05-17T23:38:13Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Experiments on CIFAR-10 demonstrate improved classification performance over a range of non-identicalness, with classification accuracy improved from 30.1% to 76.9% in the most skewed settings."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The synthetic non-identical datasets created by the authors accurately capture the statistical heterogeneity present in real-world federated visual data collected from mobile devices."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Non-identical data distributions degrade federated averaging accuracy on visual classification, but server momentum raises CIFAR-10 accuracy from 30.1% to 76.9% in the most skewed regimes."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Non-identical data distributions degrade federated averaging performance on visual tasks, but server momentum recovers most of the accuracy loss."}],"snapshot_sha256":"d6591ad3cd2b5c79480a8529ec8a44c8015e707520b670e4ef3ac7750f654ef6"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"aad5c54d1e883eb2db41b17bbc2eb572fa9028591bfb11511d29f679aa60ca68"},"paper":{"abstract_excerpt":"Federated Learning enables visual models to be trained in a privacy-preserving way using real-world data from mobile devices. Given their distributed nature, the statistics of the data across these devices is likely to differ significantly. In this work, we look at the effect such non-identical data distributions has on visual classification via Federated Learning. We propose a way to synthesize datasets with a continuous range of identicalness and provide performance measures for the Federated Averaging algorithm. We show that performance degrades as distributions differ more, and propose a m","authors_text":"Hang Qi, Matthew Brown, Tzu-Ming Harry Hsu","cross_cats":["cs.CV","stat.ML"],"headline":"Non-identical data distributions degrade federated averaging performance on visual tasks, but server momentum recovers most of the accuracy loss.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-09-13T17:26:20Z","title":"Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification"},"references":{"count":20,"internal_anchors":4,"resolved_work":20,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Learning multiple layers of features from tiny images","work_id":"533d9d88-54d2-4699-9f13-27ab1746339c","year":2009},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"International Conference on Learning Representations , year =","work_id":"74e5da63-cffe-4f1a-a090-e0b98d805e10","year":1907},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Communication-efficient learning of deep networks from decentralized data","work_id":"5c6a6545-af99-43fb-8083-08ceaf806562","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":6,"title":"Gradient methods for minimizing composite objective function","work_id":"82d58aed-7045-4ac7-af05-ef90cac9ee7a","year":2007},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":10,"title":"Advanced convolutional neural networks","work_id":"d468776e-70d7-4ba4-8062-11555e000593","year":null}],"snapshot_sha256":"a08ef1b4d83b678b3b7b3048f52054dcdb4ffab61ea2f38ff89eb0afdab3535f"},"source":{"id":"1909.06335","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-17T17:31:04.648279Z","id":"fcaa14a9-8382-4729-9e94-dcfe37dd012a","model_set":{"reader":"grok-4.3"},"one_line_summary":"Non-identical data distributions degrade federated averaging accuracy on visual classification, but server momentum raises CIFAR-10 accuracy from 30.1% to 76.9% in the most skewed regimes.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Non-identical data distributions degrade federated averaging performance on visual tasks, but server momentum recovers most of the accuracy loss.","strongest_claim":"Experiments on CIFAR-10 demonstrate improved classification performance over a range of non-identicalness, with classification accuracy improved from 30.1% to 76.9% in the most skewed settings.","weakest_assumption":"The synthetic non-identical datasets created by the authors accurately capture the statistical heterogeneity present in real-world federated visual data collected from mobile devices."}},"verdict_id":"fcaa14a9-8382-4729-9e94-dcfe37dd012a"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:100dfe4e051a53b2f775bdb9806c40a7c4e9d582893183203bd0ab9fd0da1ea2","target":"record","created_at":"2026-05-17T23:38:13Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"ea16202aaf1c40a1a276d37b9147b273854430fed81d2d85e909613bdca48229","cross_cats_sorted":["cs.CV","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-09-13T17:26:20Z","title_canon_sha256":"bdf8dc665285f408ac522b864cb845415d7e8dfc30ccbca9450d38eaf2efea69"},"schema_version":"1.0","source":{"id":"1909.06335","kind":"arxiv","version":1}},"canonical_sha256":"ae41908d02ff7549b09904990b43e908350564c0858783c17d226d1d1b63b6ae","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ae41908d02ff7549b09904990b43e908350564c0858783c17d226d1d1b63b6ae","first_computed_at":"2026-05-17T23:38:13.504555Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:13.504555Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"yWhnESu+nQDqo9vEj5Zr4qeoTHFYfiiAbYfsrbD/XGYdtxPcDOGUwVYNnC90MZE9oBuKxkVyuN4UNK6YiJc4Ag==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:13.505205Z","signed_message":"canonical_sha256_bytes"},"source_id":"1909.06335","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:100dfe4e051a53b2f775bdb9806c40a7c4e9d582893183203bd0ab9fd0da1ea2","sha256:ecf46b04836f560b3a6c3bdc296e5560b763a5088bf77027347c6a269ea6a807"],"state_sha256":"2735cd0eb32d4b762032cd3364fe47197e2bc7a57cd4cee46ab8764663fa29d0"}