{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6AQZUDFJIBQ3RKK4VCAD2BSWU3","short_pith_number":"pith:6AQZUDFJ","schema_version":"1.0","canonical_sha256":"f0219a0ca94061b8a95ca8803d0656a6fad9c1ebec2485b402b15f07b5cafd56","source":{"kind":"arxiv","id":"2411.17567","version":1},"attestation_state":"computed","paper":{"title":"Improving the Convergence Rates of Forward Gradient Descent with Repeated Sampling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.NE","stat.TH"],"primary_cat":"math.ST","authors_text":"Johannes Schmidt-Hieber, Niklas Dexheimer","submitted_at":"2024-11-26T16:28:16Z","abstract_excerpt":"Forward gradient descent (FGD) has been proposed as a biologically more plausible alternative of gradient descent as it can be computed without backward pass. Considering the linear model with $d$ parameters, previous work has found that the prediction error of FGD is, however, by a factor $d$ slower than the prediction error of stochastic gradient descent (SGD). In this paper we show that by computing $\\ell$ FGD steps based on each training sample, this suboptimality factor becomes $d/(\\ell \\wedge d)$ and thus the suboptimality of the rate disappears if $\\ell \\gtrsim d.$ We also show that FGD"},"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":"2411.17567","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"math.ST","submitted_at":"2024-11-26T16:28:16Z","cross_cats_sorted":["cs.LG","cs.NE","stat.TH"],"title_canon_sha256":"2457f0c4056de7f1cf9883dc617d11e1d69040b63783a995e5712f3b4eebec68","abstract_canon_sha256":"9ba7ea03eb4e92946cb25b179f9b714d78ca6bedcd46634032b0ad2458f99792"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:40:49.095772Z","signature_b64":"4wfhiZ9ya8It1Bw/X2pZg4ZST3El+nYUyahD32BaBPL86PUIWEu34pyfKiICn/iR1Bdg/PaQxcHHKhKtWoisBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f0219a0ca94061b8a95ca8803d0656a6fad9c1ebec2485b402b15f07b5cafd56","last_reissued_at":"2026-07-05T09:40:49.095219Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:40:49.095219Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improving the Convergence Rates of Forward Gradient Descent with Repeated Sampling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.NE","stat.TH"],"primary_cat":"math.ST","authors_text":"Johannes Schmidt-Hieber, Niklas Dexheimer","submitted_at":"2024-11-26T16:28:16Z","abstract_excerpt":"Forward gradient descent (FGD) has been proposed as a biologically more plausible alternative of gradient descent as it can be computed without backward pass. Considering the linear model with $d$ parameters, previous work has found that the prediction error of FGD is, however, by a factor $d$ slower than the prediction error of stochastic gradient descent (SGD). In this paper we show that by computing $\\ell$ FGD steps based on each training sample, this suboptimality factor becomes $d/(\\ell \\wedge d)$ and thus the suboptimality of the rate disappears if $\\ell \\gtrsim d.$ We also show that FGD"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.17567","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/2411.17567/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":"2411.17567","created_at":"2026-07-05T09:40:49.095287+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.17567v1","created_at":"2026-07-05T09:40:49.095287+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.17567","created_at":"2026-07-05T09:40:49.095287+00:00"},{"alias_kind":"pith_short_12","alias_value":"6AQZUDFJIBQ3","created_at":"2026-07-05T09:40:49.095287+00:00"},{"alias_kind":"pith_short_16","alias_value":"6AQZUDFJIBQ3RKK4","created_at":"2026-07-05T09:40:49.095287+00:00"},{"alias_kind":"pith_short_8","alias_value":"6AQZUDFJ","created_at":"2026-07-05T09:40:49.095287+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09734","citing_title":"Adaptive directional gradients for parameterised quantum circuits","ref_index":58,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3","json":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3.json","graph_json":"https://pith.science/api/pith-number/6AQZUDFJIBQ3RKK4VCAD2BSWU3/graph.json","events_json":"https://pith.science/api/pith-number/6AQZUDFJIBQ3RKK4VCAD2BSWU3/events.json","paper":"https://pith.science/paper/6AQZUDFJ"},"agent_actions":{"view_html":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3","download_json":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3.json","view_paper":"https://pith.science/paper/6AQZUDFJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.17567&json=true","fetch_graph":"https://pith.science/api/pith-number/6AQZUDFJIBQ3RKK4VCAD2BSWU3/graph.json","fetch_events":"https://pith.science/api/pith-number/6AQZUDFJIBQ3RKK4VCAD2BSWU3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3/action/storage_attestation","attest_author":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3/action/author_attestation","sign_citation":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3/action/citation_signature","submit_replication":"https://pith.science/pith/6AQZUDFJIBQ3RKK4VCAD2BSWU3/action/replication_record"}},"created_at":"2026-07-05T09:40:49.095287+00:00","updated_at":"2026-07-05T09:40:49.095287+00:00"}