{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:VZLWTYT2RSIKBOLV4ZHBDH72U4","short_pith_number":"pith:VZLWTYT2","schema_version":"1.0","canonical_sha256":"ae5769e27a8c90a0b975e64e119ffaa71e8efa6944a8a6d49463b72b81fc9690","source":{"kind":"arxiv","id":"2202.08587","version":1},"attestation_state":"computed","paper":{"title":"Gradients without Backpropagation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"At{\\i}l{\\i}m G\\\"une\\c{s} Baydin, Barak A. Pearlmutter, Don Syme, Frank Wood, Philip Torr","submitted_at":"2022-02-17T11:07:55Z","abstract_excerpt":"Using backpropagation to compute gradients of objective functions for optimization has remained a mainstay of machine learning. Backpropagation, or reverse-mode differentiation, is a special case within the general family of automatic differentiation algorithms that also includes the forward mode. We present a method to compute gradients based solely on the directional derivative that one can compute exactly and efficiently via the forward mode. We call this formulation the forward gradient, an unbiased estimate of the gradient that can be evaluated in a single forward run of the function, ent"},"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":"2202.08587","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-02-17T11:07:55Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"45173bcb90c9717f046c0c57bb499dd42993e4df329092616c6a2a4c0eb8dba3","abstract_canon_sha256":"6d689532ec02a234ef46ac9b3275f9e30ea467301a14a566d316c86023c4ab17"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:57:55.942873Z","signature_b64":"8E8+UqTViqUklX1qLMq82+iwqsuJZaRhRRJmz6w/swN3sEIviAY7KMbcsFN9nBX+cYOfMG5iBrwvPMiFJwNJCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ae5769e27a8c90a0b975e64e119ffaa71e8efa6944a8a6d49463b72b81fc9690","last_reissued_at":"2026-07-05T03:57:55.942475Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:57:55.942475Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradients without Backpropagation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"At{\\i}l{\\i}m G\\\"une\\c{s} Baydin, Barak A. Pearlmutter, Don Syme, Frank Wood, Philip Torr","submitted_at":"2022-02-17T11:07:55Z","abstract_excerpt":"Using backpropagation to compute gradients of objective functions for optimization has remained a mainstay of machine learning. Backpropagation, or reverse-mode differentiation, is a special case within the general family of automatic differentiation algorithms that also includes the forward mode. We present a method to compute gradients based solely on the directional derivative that one can compute exactly and efficiently via the forward mode. We call this formulation the forward gradient, an unbiased estimate of the gradient that can be evaluated in a single forward run of the function, ent"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.08587","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/2202.08587/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":"2202.08587","created_at":"2026-07-05T03:57:55.942532+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.08587v1","created_at":"2026-07-05T03:57:55.942532+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.08587","created_at":"2026-07-05T03:57:55.942532+00:00"},{"alias_kind":"pith_short_12","alias_value":"VZLWTYT2RSIK","created_at":"2026-07-05T03:57:55.942532+00:00"},{"alias_kind":"pith_short_16","alias_value":"VZLWTYT2RSIKBOLV","created_at":"2026-07-05T03:57:55.942532+00:00"},{"alias_kind":"pith_short_8","alias_value":"VZLWTYT2","created_at":"2026-07-05T03:57:55.942532+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09734","citing_title":"Adaptive directional gradients for parameterised quantum circuits","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15622","citing_title":"Position: Zeroth-Order Optimization in Deep Learning Is Underexplored, Not Underpowered","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2603.16437","citing_title":"Dimensional Type Systems and Deterministic Memory Management: Design-Time Semantic Preservation in Native Compilation","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2603.18104","citing_title":"Adaptive Domain Models: Bayesian Evolution, Warm Rotation, and Principled Training for Geometric and Neuromorphic AI","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2603.17627","citing_title":"The Program Hypergraph: Multi-Way Relational Structure for Geometric Algebra, Spatial Compute, and Physics-Aware Compilation","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2603.25414","citing_title":"Decidable By Construction: Design-Time Verification for Trustworthy AI","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00740","citing_title":"Randomized Subspace Nesterov Accelerated Gradient","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01928","citing_title":"Training Non-Differentiable Networks via Optimal Transport","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08204","citing_title":"Introducing Echo Networks for Computational Neuroevolution","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4","json":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4.json","graph_json":"https://pith.science/api/pith-number/VZLWTYT2RSIKBOLV4ZHBDH72U4/graph.json","events_json":"https://pith.science/api/pith-number/VZLWTYT2RSIKBOLV4ZHBDH72U4/events.json","paper":"https://pith.science/paper/VZLWTYT2"},"agent_actions":{"view_html":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4","download_json":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4.json","view_paper":"https://pith.science/paper/VZLWTYT2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.08587&json=true","fetch_graph":"https://pith.science/api/pith-number/VZLWTYT2RSIKBOLV4ZHBDH72U4/graph.json","fetch_events":"https://pith.science/api/pith-number/VZLWTYT2RSIKBOLV4ZHBDH72U4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4/action/storage_attestation","attest_author":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4/action/author_attestation","sign_citation":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4/action/citation_signature","submit_replication":"https://pith.science/pith/VZLWTYT2RSIKBOLV4ZHBDH72U4/action/replication_record"}},"created_at":"2026-07-05T03:57:55.942532+00:00","updated_at":"2026-07-05T03:57:55.942532+00:00"}