{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2012:M3HLQ7XSR3AJOTIRN2BM4X4JJS","short_pith_number":"pith:M3HLQ7XS","schema_version":"1.0","canonical_sha256":"66ceb87ef28ec0974d116e82ce5f894c8dcae064b7efd3503d3bb8d530f355d2","source":{"kind":"arxiv","id":"1206.6464","version":2},"attestation_state":"computed","paper":{"title":"Estimating the Hessian by Back-propagating Curvature","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ilya Sutskever (University of Toronto), James Martens (University of Toronto), Kevin Swersky (University of Toronto)","submitted_at":"2012-06-27T19:59:59Z","abstract_excerpt":"In this work we develop Curvature Propagation (CP), a general technique for efficiently computing unbiased approximations of the Hessian of any function that is computed using a computational graph. At the cost of roughly two gradient evaluations, CP can give a rank-1 approximation of the whole Hessian, and can be repeatedly applied to give increasingly precise unbiased estimates of any or all of the entries of the Hessian. Of particular interest is the diagonal of the Hessian, for which no general approach is known to exist that is both efficient and accurate. We show in experiments that CP t"},"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":"1206.6464","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2012-06-27T19:59:59Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"efef928d3ce711bc5da57398bb12394b1c5e0a67345ab352647ac9a7158f6358","abstract_canon_sha256":"1fad6a817e422ae2eccbb3bc5b718d5dfa352ed7519eac0dab5dcd4e735092e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:46:18.259448Z","signature_b64":"mhV4cOSQExdS7TD9zk+bIP+0ZRUr2qq32ZsC/XuhHAdUISg8RWhZmC+sgVm3t6DEqBuooPk9VteuIGKvXCdcAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"66ceb87ef28ec0974d116e82ce5f894c8dcae064b7efd3503d3bb8d530f355d2","last_reissued_at":"2026-05-18T03:46:18.258747Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:46:18.258747Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Estimating the Hessian by Back-propagating Curvature","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Ilya Sutskever (University of Toronto), James Martens (University of Toronto), Kevin Swersky (University of Toronto)","submitted_at":"2012-06-27T19:59:59Z","abstract_excerpt":"In this work we develop Curvature Propagation (CP), a general technique for efficiently computing unbiased approximations of the Hessian of any function that is computed using a computational graph. At the cost of roughly two gradient evaluations, CP can give a rank-1 approximation of the whole Hessian, and can be repeatedly applied to give increasingly precise unbiased estimates of any or all of the entries of the Hessian. Of particular interest is the diagonal of the Hessian, for which no general approach is known to exist that is both efficient and accurate. We show in experiments that CP t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1206.6464","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"1206.6464","created_at":"2026-05-18T03:46:18.258873+00:00"},{"alias_kind":"arxiv_version","alias_value":"1206.6464v2","created_at":"2026-05-18T03:46:18.258873+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1206.6464","created_at":"2026-05-18T03:46:18.258873+00:00"},{"alias_kind":"pith_short_12","alias_value":"M3HLQ7XSR3AJ","created_at":"2026-05-18T12:27:14.488303+00:00"},{"alias_kind":"pith_short_16","alias_value":"M3HLQ7XSR3AJOTIR","created_at":"2026-05-18T12:27:14.488303+00:00"},{"alias_kind":"pith_short_8","alias_value":"M3HLQ7XS","created_at":"2026-05-18T12:27:14.488303+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.13771","citing_title":"Score-Based Training for Energy-Based TTS Models","ref_index":36,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS","json":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS.json","graph_json":"https://pith.science/api/pith-number/M3HLQ7XSR3AJOTIRN2BM4X4JJS/graph.json","events_json":"https://pith.science/api/pith-number/M3HLQ7XSR3AJOTIRN2BM4X4JJS/events.json","paper":"https://pith.science/paper/M3HLQ7XS"},"agent_actions":{"view_html":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS","download_json":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS.json","view_paper":"https://pith.science/paper/M3HLQ7XS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1206.6464&json=true","fetch_graph":"https://pith.science/api/pith-number/M3HLQ7XSR3AJOTIRN2BM4X4JJS/graph.json","fetch_events":"https://pith.science/api/pith-number/M3HLQ7XSR3AJOTIRN2BM4X4JJS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS/action/storage_attestation","attest_author":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS/action/author_attestation","sign_citation":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS/action/citation_signature","submit_replication":"https://pith.science/pith/M3HLQ7XSR3AJOTIRN2BM4X4JJS/action/replication_record"}},"created_at":"2026-05-18T03:46:18.258873+00:00","updated_at":"2026-05-18T03:46:18.258873+00:00"}