{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:4YKGQG43MFT4MN7PMVMS4XHOYR","short_pith_number":"pith:4YKGQG43","schema_version":"1.0","canonical_sha256":"e614681b9b6167c637ef65592e5ceec46bb944dfb5d058fae15675d479314f94","source":{"kind":"arxiv","id":"2003.00307","version":2},"attestation_state":"computed","paper":{"title":"Loss landscapes and optimization in over-parameterized non-linear systems and neural networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Chaoyue Liu, Libin Zhu, Mikhail Belkin","submitted_at":"2020-02-29T17:18:28Z","abstract_excerpt":"The success of deep learning is due, to a large extent, to the remarkable effectiveness of gradient-based optimization methods applied to large neural networks. The purpose of this work is to propose a modern view and a general mathematical framework for loss landscapes and efficient optimization in over-parameterized machine learning models and systems of non-linear equations, a setting that includes over-parameterized deep neural networks. Our starting observation is that optimization problems corresponding to such systems are generally not convex, even locally. We argue that instead they sa"},"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":"2003.00307","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-02-29T17:18:28Z","cross_cats_sorted":["math.OC","stat.ML"],"title_canon_sha256":"1f0fd35161430b1af4da7bb4493478eace834a28ea68d9eea0b3cfa57913fc5b","abstract_canon_sha256":"43e110499885ca8b700a25c58769b568577bfe46ffe559bd1e97a52812afcdcb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:43:42.584856Z","signature_b64":"bsPe9wVdXAmxP1RtrDEcjkOeFeOc7mCdl+lP4dd2hpZ+1VkqRkOSHN/ia0lCVi45dXP35un0te1+ff9OpTAlDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e614681b9b6167c637ef65592e5ceec46bb944dfb5d058fae15675d479314f94","last_reissued_at":"2026-07-05T02:43:42.584350Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:43:42.584350Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Loss landscapes and optimization in over-parameterized non-linear systems and neural networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Chaoyue Liu, Libin Zhu, Mikhail Belkin","submitted_at":"2020-02-29T17:18:28Z","abstract_excerpt":"The success of deep learning is due, to a large extent, to the remarkable effectiveness of gradient-based optimization methods applied to large neural networks. The purpose of this work is to propose a modern view and a general mathematical framework for loss landscapes and efficient optimization in over-parameterized machine learning models and systems of non-linear equations, a setting that includes over-parameterized deep neural networks. Our starting observation is that optimization problems corresponding to such systems are generally not convex, even locally. We argue that instead they sa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2003.00307","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2003.00307/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":"2003.00307","created_at":"2026-07-05T02:43:42.584414+00:00"},{"alias_kind":"arxiv_version","alias_value":"2003.00307v2","created_at":"2026-07-05T02:43:42.584414+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2003.00307","created_at":"2026-07-05T02:43:42.584414+00:00"},{"alias_kind":"pith_short_12","alias_value":"4YKGQG43MFT4","created_at":"2026-07-05T02:43:42.584414+00:00"},{"alias_kind":"pith_short_16","alias_value":"4YKGQG43MFT4MN7P","created_at":"2026-07-05T02:43:42.584414+00:00"},{"alias_kind":"pith_short_8","alias_value":"4YKGQG43","created_at":"2026-07-05T02:43:42.584414+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10089","citing_title":"A Theory on Flow Matching with Neural Networks","ref_index":269,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24908","citing_title":"On the Impact of Class Imbalance on the Learning Dynamics of Deep Neural Networks:An Intuitive Insight","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2401.01335","citing_title":"Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models","ref_index":251,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR","json":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR.json","graph_json":"https://pith.science/api/pith-number/4YKGQG43MFT4MN7PMVMS4XHOYR/graph.json","events_json":"https://pith.science/api/pith-number/4YKGQG43MFT4MN7PMVMS4XHOYR/events.json","paper":"https://pith.science/paper/4YKGQG43"},"agent_actions":{"view_html":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR","download_json":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR.json","view_paper":"https://pith.science/paper/4YKGQG43","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2003.00307&json=true","fetch_graph":"https://pith.science/api/pith-number/4YKGQG43MFT4MN7PMVMS4XHOYR/graph.json","fetch_events":"https://pith.science/api/pith-number/4YKGQG43MFT4MN7PMVMS4XHOYR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR/action/storage_attestation","attest_author":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR/action/author_attestation","sign_citation":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR/action/citation_signature","submit_replication":"https://pith.science/pith/4YKGQG43MFT4MN7PMVMS4XHOYR/action/replication_record"}},"created_at":"2026-07-05T02:43:42.584414+00:00","updated_at":"2026-07-05T02:43:42.584414+00:00"}