{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:K7T63IVCGEW3YKH2LSIU32UNJV","short_pith_number":"pith:K7T63IVC","schema_version":"1.0","canonical_sha256":"57e7eda2a2312dbc28fa5c914dea8d4d73222fcdae9a77ecb7f9b73f51fd406f","source":{"kind":"arxiv","id":"2012.13632","version":1},"attestation_state":"computed","paper":{"title":"Adaptively Solving the Local-Minimum Problem for Deep Neural Networks","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Huachuan Wang, James Ting-Ho Lo","submitted_at":"2020-12-25T21:51:48Z","abstract_excerpt":"This paper aims to overcome a fundamental problem in the theory and application of deep neural networks (DNNs). We propose a method to solve the local minimum problem in training DNNs directly. Our method is based on the cross-entropy loss criterion's convexification by transforming the cross-entropy loss into a risk averting error (RAE) criterion. To alleviate numerical difficulties, a normalized RAE (NRAE) is employed. The convexity region of the cross-entropy loss expands as its risk sensitivity index (RSI) increases. Making the best use of the convexity region, our method starts training w"},"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":"2012.13632","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2020-12-25T21:51:48Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"c0d6df4003452f27694bb0979c46e30931d35e6cf41f998b533c57da7fafde58","abstract_canon_sha256":"3e9a8cd2a7e700c3ce9f6223e22426c22595b6aee7c232c7d98e21a427f3f2a3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:02:11.605536Z","signature_b64":"sMLtsJEuus8SnKhehcyqfKFjLTM18od4ziM9YsF9pM+6cin24OfMaB3reoxzO4nlkM0jU5wtFSoX9qhF9wvHCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"57e7eda2a2312dbc28fa5c914dea8d4d73222fcdae9a77ecb7f9b73f51fd406f","last_reissued_at":"2026-07-05T02:02:11.605114Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:02:11.605114Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adaptively Solving the Local-Minimum Problem for Deep Neural Networks","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Huachuan Wang, James Ting-Ho Lo","submitted_at":"2020-12-25T21:51:48Z","abstract_excerpt":"This paper aims to overcome a fundamental problem in the theory and application of deep neural networks (DNNs). We propose a method to solve the local minimum problem in training DNNs directly. Our method is based on the cross-entropy loss criterion's convexification by transforming the cross-entropy loss into a risk averting error (RAE) criterion. To alleviate numerical difficulties, a normalized RAE (NRAE) is employed. The convexity region of the cross-entropy loss expands as its risk sensitivity index (RSI) increases. Making the best use of the convexity region, our method starts training w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.13632","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/2012.13632/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":"2012.13632","created_at":"2026-07-05T02:02:11.605172+00:00"},{"alias_kind":"arxiv_version","alias_value":"2012.13632v1","created_at":"2026-07-05T02:02:11.605172+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.13632","created_at":"2026-07-05T02:02:11.605172+00:00"},{"alias_kind":"pith_short_12","alias_value":"K7T63IVCGEW3","created_at":"2026-07-05T02:02:11.605172+00:00"},{"alias_kind":"pith_short_16","alias_value":"K7T63IVCGEW3YKH2","created_at":"2026-07-05T02:02:11.605172+00:00"},{"alias_kind":"pith_short_8","alias_value":"K7T63IVC","created_at":"2026-07-05T02:02:11.605172+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.10649","citing_title":"Deep Loss Convexification for Learning Iterative Models","ref_index":33,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV","json":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV.json","graph_json":"https://pith.science/api/pith-number/K7T63IVCGEW3YKH2LSIU32UNJV/graph.json","events_json":"https://pith.science/api/pith-number/K7T63IVCGEW3YKH2LSIU32UNJV/events.json","paper":"https://pith.science/paper/K7T63IVC"},"agent_actions":{"view_html":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV","download_json":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV.json","view_paper":"https://pith.science/paper/K7T63IVC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2012.13632&json=true","fetch_graph":"https://pith.science/api/pith-number/K7T63IVCGEW3YKH2LSIU32UNJV/graph.json","fetch_events":"https://pith.science/api/pith-number/K7T63IVCGEW3YKH2LSIU32UNJV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV/action/storage_attestation","attest_author":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV/action/author_attestation","sign_citation":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV/action/citation_signature","submit_replication":"https://pith.science/pith/K7T63IVCGEW3YKH2LSIU32UNJV/action/replication_record"}},"created_at":"2026-07-05T02:02:11.605172+00:00","updated_at":"2026-07-05T02:02:11.605172+00:00"}