{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:MHSCAI5YPBPJPNTRTUIHZUXB3U","short_pith_number":"pith:MHSCAI5Y","schema_version":"1.0","canonical_sha256":"61e42023b8785e97b6719d107cd2e1dd3d0debd8c529523c30b48c51731e5df2","source":{"kind":"arxiv","id":"2210.03310","version":3},"attestation_state":"computed","paper":{"title":"Scaling Forward Gradient With Local Losses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV","cs.NE"],"primary_cat":"cs.LG","authors_text":"Geoffrey Hinton, Mengye Ren, Renjie Liao, Simon Kornblith","submitted_at":"2022-10-07T03:52:27Z","abstract_excerpt":"Forward gradient learning computes a noisy directional gradient and is a biologically plausible alternative to backprop for learning deep neural networks. However, the standard forward gradient algorithm, when applied naively, suffers from high variance when the number of parameters to be learned is large. In this paper, we propose a series of architectural and algorithmic modifications that together make forward gradient learning practical for standard deep learning benchmark tasks. We show that it is possible to substantially reduce the variance of the forward gradient estimator by applying "},"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":"2210.03310","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-10-07T03:52:27Z","cross_cats_sorted":["cs.CV","cs.NE"],"title_canon_sha256":"92f7485a286247ce13425e5d803a99d306eac0d69fe1e3e7dec809b7bbd86783","abstract_canon_sha256":"0632a613643bf0423659089330ff2b8e4afa08eb153ab850c39a82b917651c30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:47:18.119227Z","signature_b64":"4TmUfjokwvEZ7KhB1Zx7odroirVPE7mu4MgV5BPRcCgmPu5jznCih6bGou1GbQVdlUuiO0bDPjv0CTtyhBo9CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"61e42023b8785e97b6719d107cd2e1dd3d0debd8c529523c30b48c51731e5df2","last_reissued_at":"2026-07-05T05:47:18.118730Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:47:18.118730Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Forward Gradient With Local Losses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV","cs.NE"],"primary_cat":"cs.LG","authors_text":"Geoffrey Hinton, Mengye Ren, Renjie Liao, Simon Kornblith","submitted_at":"2022-10-07T03:52:27Z","abstract_excerpt":"Forward gradient learning computes a noisy directional gradient and is a biologically plausible alternative to backprop for learning deep neural networks. However, the standard forward gradient algorithm, when applied naively, suffers from high variance when the number of parameters to be learned is large. In this paper, we propose a series of architectural and algorithmic modifications that together make forward gradient learning practical for standard deep learning benchmark tasks. We show that it is possible to substantially reduce the variance of the forward gradient estimator by applying "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.03310","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/2210.03310/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":"2210.03310","created_at":"2026-07-05T05:47:18.118790+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.03310v3","created_at":"2026-07-05T05:47:18.118790+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.03310","created_at":"2026-07-05T05:47:18.118790+00:00"},{"alias_kind":"pith_short_12","alias_value":"MHSCAI5YPBPJ","created_at":"2026-07-05T05:47:18.118790+00:00"},{"alias_kind":"pith_short_16","alias_value":"MHSCAI5YPBPJPNTR","created_at":"2026-07-05T05:47:18.118790+00:00"},{"alias_kind":"pith_short_8","alias_value":"MHSCAI5Y","created_at":"2026-07-05T05:47:18.118790+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09734","citing_title":"Adaptive directional gradients for parameterised quantum circuits","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19533","citing_title":"Replacement Learning: Training Neural Networks with Fewer Parameters","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U","json":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U.json","graph_json":"https://pith.science/api/pith-number/MHSCAI5YPBPJPNTRTUIHZUXB3U/graph.json","events_json":"https://pith.science/api/pith-number/MHSCAI5YPBPJPNTRTUIHZUXB3U/events.json","paper":"https://pith.science/paper/MHSCAI5Y"},"agent_actions":{"view_html":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U","download_json":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U.json","view_paper":"https://pith.science/paper/MHSCAI5Y","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.03310&json=true","fetch_graph":"https://pith.science/api/pith-number/MHSCAI5YPBPJPNTRTUIHZUXB3U/graph.json","fetch_events":"https://pith.science/api/pith-number/MHSCAI5YPBPJPNTRTUIHZUXB3U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U/action/storage_attestation","attest_author":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U/action/author_attestation","sign_citation":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U/action/citation_signature","submit_replication":"https://pith.science/pith/MHSCAI5YPBPJPNTRTUIHZUXB3U/action/replication_record"}},"created_at":"2026-07-05T05:47:18.118790+00:00","updated_at":"2026-07-05T05:47:18.118790+00:00"}