{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2013:7HT45CUO6TCD5Q2CXOBQENHTJU","short_pith_number":"pith:7HT45CUO","schema_version":"1.0","canonical_sha256":"f9e7ce8a8ef4c43ec342bb830234f34d1adb6d7ea7182c7d3896990c20523870","source":{"kind":"arxiv","id":"1307.1493","version":2},"attestation_state":"computed","paper":{"title":"Dropout Training as Adaptive Regularization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ME"],"primary_cat":"stat.ML","authors_text":"Percy Liang, Sida Wang, Stefan Wager","submitted_at":"2013-07-04T21:33:56Z","abstract_excerpt":"Dropout and other feature noising schemes control overfitting by artificially corrupting the training data. For generalized linear models, dropout performs a form of adaptive regularization. Using this viewpoint, we show that the dropout regularizer is first-order equivalent to an L2 regularizer applied after scaling the features by an estimate of the inverse diagonal Fisher information matrix. We also establish a connection to AdaGrad, an online learning algorithm, and find that a close relative of AdaGrad operates by repeatedly solving linear dropout-regularized problems. By casting dropout "},"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":"1307.1493","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2013-07-04T21:33:56Z","cross_cats_sorted":["cs.LG","stat.ME"],"title_canon_sha256":"165d2fd03c1d90d401afda0d5f7d13f9d8f3808564e0c1f6196f6f27953c6c1a","abstract_canon_sha256":"40e5504dd3ec2f3573c8b5552d5f478455d8a1e569f29dfa03fd40ce6605e685"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:08:15.786451Z","signature_b64":"EKbjy/20NHXUw/xUx+gLaOoJk2UaxKc5jYg76k/Hg9aZWY2/cxBKAFrPPU36JpGM3kZXCqKoi6Ykq6Pd1wYOCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9e7ce8a8ef4c43ec342bb830234f34d1adb6d7ea7182c7d3896990c20523870","last_reissued_at":"2026-05-18T03:08:15.785699Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:08:15.785699Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dropout Training as Adaptive Regularization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ME"],"primary_cat":"stat.ML","authors_text":"Percy Liang, Sida Wang, Stefan Wager","submitted_at":"2013-07-04T21:33:56Z","abstract_excerpt":"Dropout and other feature noising schemes control overfitting by artificially corrupting the training data. For generalized linear models, dropout performs a form of adaptive regularization. Using this viewpoint, we show that the dropout regularizer is first-order equivalent to an L2 regularizer applied after scaling the features by an estimate of the inverse diagonal Fisher information matrix. We also establish a connection to AdaGrad, an online learning algorithm, and find that a close relative of AdaGrad operates by repeatedly solving linear dropout-regularized problems. By casting dropout "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1307.1493","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":"1307.1493","created_at":"2026-05-18T03:08:15.785807+00:00"},{"alias_kind":"arxiv_version","alias_value":"1307.1493v2","created_at":"2026-05-18T03:08:15.785807+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1307.1493","created_at":"2026-05-18T03:08:15.785807+00:00"},{"alias_kind":"pith_short_12","alias_value":"7HT45CUO6TCD","created_at":"2026-05-18T12:27:36.564083+00:00"},{"alias_kind":"pith_short_16","alias_value":"7HT45CUO6TCD5Q2C","created_at":"2026-05-18T12:27:36.564083+00:00"},{"alias_kind":"pith_short_8","alias_value":"7HT45CUO","created_at":"2026-05-18T12:27:36.564083+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.05436","citing_title":"Estimating Implicit Regularization in Deep Learning","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU","json":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU.json","graph_json":"https://pith.science/api/pith-number/7HT45CUO6TCD5Q2CXOBQENHTJU/graph.json","events_json":"https://pith.science/api/pith-number/7HT45CUO6TCD5Q2CXOBQENHTJU/events.json","paper":"https://pith.science/paper/7HT45CUO"},"agent_actions":{"view_html":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU","download_json":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU.json","view_paper":"https://pith.science/paper/7HT45CUO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1307.1493&json=true","fetch_graph":"https://pith.science/api/pith-number/7HT45CUO6TCD5Q2CXOBQENHTJU/graph.json","fetch_events":"https://pith.science/api/pith-number/7HT45CUO6TCD5Q2CXOBQENHTJU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU/action/storage_attestation","attest_author":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU/action/author_attestation","sign_citation":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU/action/citation_signature","submit_replication":"https://pith.science/pith/7HT45CUO6TCD5Q2CXOBQENHTJU/action/replication_record"}},"created_at":"2026-05-18T03:08:15.785807+00:00","updated_at":"2026-05-18T03:08:15.785807+00:00"}