{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MB47VM45YODGVZQEBYU7WAEM3T","short_pith_number":"pith:MB47VM45","schema_version":"1.0","canonical_sha256":"6079fab39dc3866ae6040e29fb008cdcddec03046b3a673109882c1096967fa9","source":{"kind":"arxiv","id":"2605.13434","version":1},"attestation_state":"computed","paper":{"title":"Rescaled Asynchronous SGD: Optimal Distributed Optimization under Data and System Heterogeneity","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Rescaling worker stepsizes by computation time fixes bias in asynchronous SGD so it converges to the true global objective.","cross_cats":["cs.DC","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ammar Mahran, Artavazd Maranjyan, Peter Richt\\'arik","submitted_at":"2026-05-13T12:27:22Z","abstract_excerpt":"Asynchronous stochastic gradient descent (ASGD) is a standard way to exploit heterogeneous compute resources in distributed learning: instead of forcing fast workers to wait for slow ones, the server updates the model whenever a gradient arrives. Vanilla ASGD applies each arriving gradient with the same weight. When local data distributions are heterogeneous, this becomes problematic: faster workers contribute more updates, and we show theoretically that the method is biased toward a frequency-weighted average of the local objectives rather than the desired global objective. Existing remedies "},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.13434","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T12:27:22Z","cross_cats_sorted":["cs.DC","math.OC","stat.ML"],"title_canon_sha256":"460975d702698bd8d60f688c91f2124e253de33f880bb3d8fe8c5ddd4ab55b90","abstract_canon_sha256":"013e29306d44dbd8adc4c101bf414a6403d2edaecc85e2292f97c5010778a9dd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:47.132478Z","signature_b64":"pUp7KbyCqdpalci+3YAPUJZ8ty7c1kO6ScloBMQEfQgUO3ugbHnCkzVqyFskvzKjyPInQ+GsU0HPDXBkH/QxDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6079fab39dc3866ae6040e29fb008cdcddec03046b3a673109882c1096967fa9","last_reissued_at":"2026-05-18T02:44:47.131927Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:47.131927Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rescaled Asynchronous SGD: Optimal Distributed Optimization under Data and System Heterogeneity","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Rescaling worker stepsizes by computation time fixes bias in asynchronous SGD so it converges to the true global objective.","cross_cats":["cs.DC","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ammar Mahran, Artavazd Maranjyan, Peter Richt\\'arik","submitted_at":"2026-05-13T12:27:22Z","abstract_excerpt":"Asynchronous stochastic gradient descent (ASGD) is a standard way to exploit heterogeneous compute resources in distributed learning: instead of forcing fast workers to wait for slow ones, the server updates the model whenever a gradient arrives. Vanilla ASGD applies each arriving gradient with the same weight. When local data distributions are heterogeneous, this becomes problematic: faster workers contribute more updates, and we show theoretically that the method is biased toward a frequency-weighted average of the local objectives rather than the desired global objective. Existing remedies "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we prove that the resulting method, Rescaled ASGD, converges to stationary points of the correct global objective in the fixed-computation model. Its time complexity matches the known lower bound in the leading term, while the effects of staleness and data heterogeneity appear only in lower-order terms.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"under smoothness and bounded heterogeneity assumptions","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Rescaled ASGD recovers convergence to the true global objective by rescaling worker stepsizes proportional to computation times, matching the known time lower bound in the leading term under non-convex smoothness and bounded heterogeneity.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Rescaling worker stepsizes by computation time fixes bias in asynchronous SGD so it converges to the true global objective.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"bed43d911b513c3ba7be443a98374a6b8d93807a454477bcce4e5a28607203ab"},"source":{"id":"2605.13434","kind":"arxiv","version":1},"verdict":{"id":"3df6c769-5d61-42f4-b85b-e2d8b0175c8e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:27:27.645909Z","strongest_claim":"we prove that the resulting method, Rescaled ASGD, converges to stationary points of the correct global objective in the fixed-computation model. Its time complexity matches the known lower bound in the leading term, while the effects of staleness and data heterogeneity appear only in lower-order terms.","one_line_summary":"Rescaled ASGD recovers convergence to the true global objective by rescaling worker stepsizes proportional to computation times, matching the known time lower bound in the leading term under non-convex smoothness and bounded heterogeneity.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"under smoothness and bounded heterogeneity assumptions","pith_extraction_headline":"Rescaling worker stepsizes by computation time fixes bias in asynchronous SGD so it converges to the true global objective."},"references":{"count":300,"sample":[{"doi":"10.1109/msp.2020.2975749","year":2020,"title":"Federated Learning: Chal- lenges, Methods, and Future Directions.IEEE Signal Processing Magazine, 37(3):50–60","work_id":"4916403b-b4d2-45a0-9053-7b8f137c878f","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.48550/arxiv.1909.06335","year":1909,"title":"Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification","work_id":"1e8d2981-bd90-4c86-8676-223498b6d816","ref_index":2,"cited_arxiv_id":"1909.06335","is_internal_anchor":true},{"doi":"10.1109/tit.2017.2736066","year":2015,"title":"arXiv.org , author =","work_id":"42d7abc2-f71a-4d63-885a-c432425462ef","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.48550/arxiv.1812.01097","year":null,"title":"Leaf: A benchmark for federated settings","work_id":"94ec8be1-6cf1-4962-994c-e762a0b9c3ee","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2012,"title":"arXiv.org , author =","work_id":"a670dcca-8345-441d-9855-f6bc678e0fbd","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":300,"snapshot_sha256":"421548bf02e07daa14540f2b965d273433a08dbfcc992d1b513da02c06f42655","internal_anchors":34},"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":"2605.13434","created_at":"2026-05-18T02:44:47.132024+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.13434v1","created_at":"2026-05-18T02:44:47.132024+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13434","created_at":"2026-05-18T02:44:47.132024+00:00"},{"alias_kind":"pith_short_12","alias_value":"MB47VM45YODG","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"MB47VM45YODGVZQE","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"MB47VM45","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.20866","citing_title":"LOSCAR-SGD: Local SGD with Communication-Computation Overlap and Delay-Corrected Sparse Model Averaging","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18174","citing_title":"Ringmaster LMO: Asynchronous Linear Minimization Oracle Momentum Method","ref_index":2,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T","json":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T.json","graph_json":"https://pith.science/api/pith-number/MB47VM45YODGVZQEBYU7WAEM3T/graph.json","events_json":"https://pith.science/api/pith-number/MB47VM45YODGVZQEBYU7WAEM3T/events.json","paper":"https://pith.science/paper/MB47VM45"},"agent_actions":{"view_html":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T","download_json":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T.json","view_paper":"https://pith.science/paper/MB47VM45","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.13434&json=true","fetch_graph":"https://pith.science/api/pith-number/MB47VM45YODGVZQEBYU7WAEM3T/graph.json","fetch_events":"https://pith.science/api/pith-number/MB47VM45YODGVZQEBYU7WAEM3T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T/action/storage_attestation","attest_author":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T/action/author_attestation","sign_citation":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T/action/citation_signature","submit_replication":"https://pith.science/pith/MB47VM45YODGVZQEBYU7WAEM3T/action/replication_record"}},"created_at":"2026-05-18T02:44:47.132024+00:00","updated_at":"2026-05-18T02:44:47.132024+00:00"}