{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6SHJGYHIABE27KYFGFGES66NLL","short_pith_number":"pith:6SHJGYHI","schema_version":"1.0","canonical_sha256":"f48e9360e80049afab05314c497bcd5ae02e2fb51fda28f7ac2282cee5a4a040","source":{"kind":"arxiv","id":"2504.20096","version":1},"attestation_state":"computed","paper":{"title":"Towards Practical Second-Order Optimizers in Deep Learning: Insights from Fisher Information Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Damien Martins Gomes","submitted_at":"2025-04-26T05:02:21Z","abstract_excerpt":"First-order optimization methods remain the standard for training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by preconditioning the stochastic gradient with a diagonal matrix. Despite the widespread adoption of first-order methods, second-order optimization algorithms often exhibit superior convergence compared to methods like Adam and SGD. However, their practicality in training DNNs is still limited by a significantly higher per-iteration computational cost compared to first-order methods. In this thesis, we present AdaFisher, a novel adaptive"},"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":"2504.20096","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-26T05:02:21Z","cross_cats_sorted":["math.OC"],"title_canon_sha256":"7ee8f870dc2032bacbb91692e1d83a50f16bfc3efc4fd0f965b066f426a06086","abstract_canon_sha256":"35ab9b5852f5596e63bedd03ca31de24babbeffd5768c326d3b12fd0f909a432"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:55:36.378284Z","signature_b64":"8e6fLb1ha3uUUaejJn9LDTsWeC54nekoWOc00by3nqApHFxRr7JU1kZo5goudIlor56C8gb8+4ihSDnZZeUdDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f48e9360e80049afab05314c497bcd5ae02e2fb51fda28f7ac2282cee5a4a040","last_reissued_at":"2026-07-05T10:55:36.377789Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:55:36.377789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Practical Second-Order Optimizers in Deep Learning: Insights from Fisher Information Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Damien Martins Gomes","submitted_at":"2025-04-26T05:02:21Z","abstract_excerpt":"First-order optimization methods remain the standard for training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by preconditioning the stochastic gradient with a diagonal matrix. Despite the widespread adoption of first-order methods, second-order optimization algorithms often exhibit superior convergence compared to methods like Adam and SGD. However, their practicality in training DNNs is still limited by a significantly higher per-iteration computational cost compared to first-order methods. In this thesis, we present AdaFisher, a novel adaptive"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.20096","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/2504.20096/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":"2504.20096","created_at":"2026-07-05T10:55:36.377842+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.20096v1","created_at":"2026-07-05T10:55:36.377842+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.20096","created_at":"2026-07-05T10:55:36.377842+00:00"},{"alias_kind":"pith_short_12","alias_value":"6SHJGYHIABE2","created_at":"2026-07-05T10:55:36.377842+00:00"},{"alias_kind":"pith_short_16","alias_value":"6SHJGYHIABE27KYF","created_at":"2026-07-05T10:55:36.377842+00:00"},{"alias_kind":"pith_short_8","alias_value":"6SHJGYHI","created_at":"2026-07-05T10:55:36.377842+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.00982","citing_title":"DHO$_2$: Accelerating Distributed Hybrid Order Optimization via Model Parallelism and ADMM","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL","json":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL.json","graph_json":"https://pith.science/api/pith-number/6SHJGYHIABE27KYFGFGES66NLL/graph.json","events_json":"https://pith.science/api/pith-number/6SHJGYHIABE27KYFGFGES66NLL/events.json","paper":"https://pith.science/paper/6SHJGYHI"},"agent_actions":{"view_html":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL","download_json":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL.json","view_paper":"https://pith.science/paper/6SHJGYHI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.20096&json=true","fetch_graph":"https://pith.science/api/pith-number/6SHJGYHIABE27KYFGFGES66NLL/graph.json","fetch_events":"https://pith.science/api/pith-number/6SHJGYHIABE27KYFGFGES66NLL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL/action/storage_attestation","attest_author":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL/action/author_attestation","sign_citation":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL/action/citation_signature","submit_replication":"https://pith.science/pith/6SHJGYHIABE27KYFGFGES66NLL/action/replication_record"}},"created_at":"2026-07-05T10:55:36.377842+00:00","updated_at":"2026-07-05T10:55:36.377842+00:00"}