{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:DLSADBD2LEEUIRAENK4OWUN7KO","short_pith_number":"pith:DLSADBD2","schema_version":"1.0","canonical_sha256":"1ae401847a59094444046ab8eb51bf539bbbf85bfe8f0382e9ad86c17e53137f","source":{"kind":"arxiv","id":"2011.02408","version":2},"attestation_state":"computed","paper":{"title":"Which Minimizer Does My Neural Network Converge To?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"David Reeb, Ingo Steinwart, Manuel Nonnenmacher","submitted_at":"2020-11-04T17:04:01Z","abstract_excerpt":"The loss surface of an overparameterized neural network (NN) possesses many global minima of zero training error. We explain how common variants of the standard NN training procedure change the minimizer obtained. First, we make explicit how the size of the initialization of a strongly overparameterized NN affects the minimizer and can deteriorate its final test performance. We propose a strategy to limit this effect. Then, we demonstrate that for adaptive optimization such as AdaGrad, the obtained minimizer generally differs from the gradient descent (GD) minimizer. This adaptive minimizer is"},"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":"2011.02408","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2020-11-04T17:04:01Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"dd810bd1fe38820d6b178e68ce8dc95c986c356ec3cab3dfbc9accd20c377f1f","abstract_canon_sha256":"f6ff4e37c518fc3e1322bb7f87f81992ab9a879b0fb0d5597b9120744097ddcb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:53:27.867240Z","signature_b64":"zkDWVWGYHJODrLKW23JrmG075RM5MPNBxjjFHBENUMiShPI6jNpOlgKTThkvt/ZKvfAzNqHkQgAaCcIe3t97Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1ae401847a59094444046ab8eb51bf539bbbf85bfe8f0382e9ad86c17e53137f","last_reissued_at":"2026-07-05T06:53:27.866750Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:53:27.866750Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Which Minimizer Does My Neural Network Converge To?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"David Reeb, Ingo Steinwart, Manuel Nonnenmacher","submitted_at":"2020-11-04T17:04:01Z","abstract_excerpt":"The loss surface of an overparameterized neural network (NN) possesses many global minima of zero training error. We explain how common variants of the standard NN training procedure change the minimizer obtained. First, we make explicit how the size of the initialization of a strongly overparameterized NN affects the minimizer and can deteriorate its final test performance. We propose a strategy to limit this effect. Then, we demonstrate that for adaptive optimization such as AdaGrad, the obtained minimizer generally differs from the gradient descent (GD) minimizer. This adaptive minimizer is"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2011.02408","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/2011.02408/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":"2011.02408","created_at":"2026-07-05T06:53:27.866819+00:00"},{"alias_kind":"arxiv_version","alias_value":"2011.02408v2","created_at":"2026-07-05T06:53:27.866819+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2011.02408","created_at":"2026-07-05T06:53:27.866819+00:00"},{"alias_kind":"pith_short_12","alias_value":"DLSADBD2LEEU","created_at":"2026-07-05T06:53:27.866819+00:00"},{"alias_kind":"pith_short_16","alias_value":"DLSADBD2LEEUIRAE","created_at":"2026-07-05T06:53:27.866819+00:00"},{"alias_kind":"pith_short_8","alias_value":"DLSADBD2","created_at":"2026-07-05T06:53:27.866819+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO","json":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO.json","graph_json":"https://pith.science/api/pith-number/DLSADBD2LEEUIRAENK4OWUN7KO/graph.json","events_json":"https://pith.science/api/pith-number/DLSADBD2LEEUIRAENK4OWUN7KO/events.json","paper":"https://pith.science/paper/DLSADBD2"},"agent_actions":{"view_html":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO","download_json":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO.json","view_paper":"https://pith.science/paper/DLSADBD2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2011.02408&json=true","fetch_graph":"https://pith.science/api/pith-number/DLSADBD2LEEUIRAENK4OWUN7KO/graph.json","fetch_events":"https://pith.science/api/pith-number/DLSADBD2LEEUIRAENK4OWUN7KO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO/action/storage_attestation","attest_author":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO/action/author_attestation","sign_citation":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO/action/citation_signature","submit_replication":"https://pith.science/pith/DLSADBD2LEEUIRAENK4OWUN7KO/action/replication_record"}},"created_at":"2026-07-05T06:53:27.866819+00:00","updated_at":"2026-07-05T06:53:27.866819+00:00"}