{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:234KR6DCQKCXTM72EWFGAAXG5V","short_pith_number":"pith:234KR6DC","canonical_record":{"source":{"id":"2012.06678","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-11T23:31:23Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"da775df861a4305faae84048f4463e76aecd6af3620932a75a861ab369a98ff6","abstract_canon_sha256":"14a98b2ef24421bfa593b9c24470b298a492a660bbbab01d932a8b622d357324"},"schema_version":"1.0"},"canonical_sha256":"d6f8a8f862828579b3fa258a6002e6ed6ab6c0eac508138209d7c435e53f440a","source":{"kind":"arxiv","id":"2012.06678","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2012.06678","created_at":"2026-05-17T23:38:46Z"},{"alias_kind":"arxiv_version","alias_value":"2012.06678v1","created_at":"2026-05-17T23:38:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.06678","created_at":"2026-05-17T23:38:46Z"},{"alias_kind":"pith_short_12","alias_value":"234KR6DCQKCX","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"234KR6DCQKCXTM72","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"234KR6DC","created_at":"2026-05-18T12:33:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:234KR6DCQKCXTM72EWFGAAXG5V","target":"record","payload":{"canonical_record":{"source":{"id":"2012.06678","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-11T23:31:23Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"da775df861a4305faae84048f4463e76aecd6af3620932a75a861ab369a98ff6","abstract_canon_sha256":"14a98b2ef24421bfa593b9c24470b298a492a660bbbab01d932a8b622d357324"},"schema_version":"1.0"},"canonical_sha256":"d6f8a8f862828579b3fa258a6002e6ed6ab6c0eac508138209d7c435e53f440a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:46.507257Z","signature_b64":"Gng8JZn3OYRs91zT2bO6cHnJuU9+FuaIWhOnpTknOKkMBhk3QSatB1DgVvkg++9iJs9kN1y/DEkvTYmwp0VICQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d6f8a8f862828579b3fa258a6002e6ed6ab6c0eac508138209d7c435e53f440a","last_reissued_at":"2026-05-17T23:38:46.506714Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:46.506714Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2012.06678","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"eKHbyVZVpSo03mWkloazfMnUfaU6L7syoLpwyHkLt18G9xsDMkTtwrNJC2yvRnReNUUH9Ach8znNfC4z4BWdCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T23:20:25.399256Z"},"content_sha256":"5ef60aa6fdf156230434f1ee8e92a08a69aef5377c590bcdacdb9d27c6fd0ea4","schema_version":"1.0","event_id":"sha256:5ef60aa6fdf156230434f1ee8e92a08a69aef5377c590bcdacdb9d27c6fd0ea4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:234KR6DCQKCXTM72EWFGAAXG5V","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"TabTransformer: Tabular Data Modeling Using Contextual Embeddings","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"TabTransformer applies self-attention to categorical feature embeddings to create contextual representations that raise prediction accuracy on tabular data.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ashish Khetan, Milan Cvitkovic, Xin Huang, Zohar Karnin","submitted_at":"2020-12-11T23:31:23Z","abstract_excerpt":"We propose TabTransformer, a novel deep tabular data modeling architecture for supervised and semi-supervised learning. The TabTransformer is built upon self-attention based Transformers. The Transformer layers transform the embeddings of categorical features into robust contextual embeddings to achieve higher prediction accuracy. Through extensive experiments on fifteen publicly available datasets, we show that the TabTransformer outperforms the state-of-the-art deep learning methods for tabular data by at least 1.0% on mean AUC, and matches the performance of tree-based ensemble models. Furt"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Through extensive experiments on fifteen publicly available datasets, we show that the TabTransformer outperforms the state-of-the-art deep learning methods for tabular data by at least 1.0% on mean AUC, and matches the performance of tree-based ensemble models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The fifteen public datasets are representative of real-world tabular distributions and that baseline deep learning and tree methods were tuned to their best possible performance without hidden advantages for the proposed model.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"TabTransformer uses Transformer self-attention to generate contextual embeddings from categorical features in tabular data, outperforming prior deep learning methods by at least 1% mean AUC and matching tree-based ensembles on 15 public datasets while showing robustness to missing and noisy features","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"TabTransformer applies self-attention to categorical feature embeddings to create contextual representations that raise prediction accuracy on tabular data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"23f5ef713f762cb055b8cadb7a1eddab47051dc174795f4dd522c783a26dbbbf"},"source":{"id":"2012.06678","kind":"arxiv","version":1},"verdict":{"id":"81d9d466-28f4-4742-8bc7-4d39225a178e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T21:28:50.990411Z","strongest_claim":"Through extensive experiments on fifteen publicly available datasets, we show that the TabTransformer outperforms the state-of-the-art deep learning methods for tabular data by at least 1.0% on mean AUC, and matches the performance of tree-based ensemble models.","one_line_summary":"TabTransformer uses Transformer self-attention to generate contextual embeddings from categorical features in tabular data, outperforming prior deep learning methods by at least 1% mean AUC and matching tree-based ensembles on 15 public datasets while showing robustness to missing and noisy features","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The fifteen public datasets are representative of real-world tabular distributions and that baseline deep learning and tree methods were tuned to their best possible performance without hidden advantages for the proposed model.","pith_extraction_headline":"TabTransformer applies self-attention to categorical feature embeddings to create contextual representations that raise prediction accuracy on tabular data."},"references":{"count":99,"sample":[{"doi":"","year":null,"title":"Proceedings of the ninth international conference on Information and knowledge management , pages=","work_id":"2a654019-2a19-4a40-8ad3-d71bd6cd4b0d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Advances in neural information processing systems , pages=","work_id":"de2dd579-e4e4-4930-ab73-1cac1aa46d25","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2002,"title":"Learning from labeled and unlabeled data with label propagation , author=. 2002 , publisher=","work_id":"af0992c3-e96a-451e-bc3e-e4a141a7dc29","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Advances in neural information processing systems , pages=","work_id":"2679e2f3-f246-4196-aac3-0b15eed5727c","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Workshop on challenges in representation learning, ICML , volume=","work_id":"eb36ae7f-1e24-491c-aef6-39195c0a348c","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":99,"snapshot_sha256":"287e376195552d6048286aa0be77469c57fb3405ac05d0c0912f26c7e9626284","internal_anchors":8},"formal_canon":{"evidence_count":1,"snapshot_sha256":"bdb99e4f0497c0418d2c49c114d6d0af1f40ecb6958b647b997902aea22e3a4d"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"81d9d466-28f4-4742-8bc7-4d39225a178e"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AuZnT8smmIDHlg8KTGfN4RnKuHOqJNgDL1QTPRnw9y4dVh8eAEUtGigsYnD80JcdNJecGWt/FWo659YOb9g4CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T23:20:25.399936Z"},"content_sha256":"c5317c9a4914b6d354a6d91b622e29de49804d0730aa6edecf310564533ea145","schema_version":"1.0","event_id":"sha256:c5317c9a4914b6d354a6d91b622e29de49804d0730aa6edecf310564533ea145"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/234KR6DCQKCXTM72EWFGAAXG5V/bundle.json","state_url":"https://pith.science/pith/234KR6DCQKCXTM72EWFGAAXG5V/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/234KR6DCQKCXTM72EWFGAAXG5V/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-27T23:20:25Z","links":{"resolver":"https://pith.science/pith/234KR6DCQKCXTM72EWFGAAXG5V","bundle":"https://pith.science/pith/234KR6DCQKCXTM72EWFGAAXG5V/bundle.json","state":"https://pith.science/pith/234KR6DCQKCXTM72EWFGAAXG5V/state.json","well_known_bundle":"https://pith.science/.well-known/pith/234KR6DCQKCXTM72EWFGAAXG5V/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:234KR6DCQKCXTM72EWFGAAXG5V","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"14a98b2ef24421bfa593b9c24470b298a492a660bbbab01d932a8b622d357324","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-11T23:31:23Z","title_canon_sha256":"da775df861a4305faae84048f4463e76aecd6af3620932a75a861ab369a98ff6"},"schema_version":"1.0","source":{"id":"2012.06678","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2012.06678","created_at":"2026-05-17T23:38:46Z"},{"alias_kind":"arxiv_version","alias_value":"2012.06678v1","created_at":"2026-05-17T23:38:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.06678","created_at":"2026-05-17T23:38:46Z"},{"alias_kind":"pith_short_12","alias_value":"234KR6DCQKCX","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"234KR6DCQKCXTM72","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"234KR6DC","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:c5317c9a4914b6d354a6d91b622e29de49804d0730aa6edecf310564533ea145","target":"graph","created_at":"2026-05-17T23:38:46Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Through extensive experiments on fifteen publicly available datasets, we show that the TabTransformer outperforms the state-of-the-art deep learning methods for tabular data by at least 1.0% on mean AUC, and matches the performance of tree-based ensemble models."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The fifteen public datasets are representative of real-world tabular distributions and that baseline deep learning and tree methods were tuned to their best possible performance without hidden advantages for the proposed model."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"TabTransformer uses Transformer self-attention to generate contextual embeddings from categorical features in tabular data, outperforming prior deep learning methods by at least 1% mean AUC and matching tree-based ensembles on 15 public datasets while showing robustness to missing and noisy features"},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"TabTransformer applies self-attention to categorical feature embeddings to create contextual representations that raise prediction accuracy on tabular data."}],"snapshot_sha256":"23f5ef713f762cb055b8cadb7a1eddab47051dc174795f4dd522c783a26dbbbf"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"bdb99e4f0497c0418d2c49c114d6d0af1f40ecb6958b647b997902aea22e3a4d"},"paper":{"abstract_excerpt":"We propose TabTransformer, a novel deep tabular data modeling architecture for supervised and semi-supervised learning. The TabTransformer is built upon self-attention based Transformers. The Transformer layers transform the embeddings of categorical features into robust contextual embeddings to achieve higher prediction accuracy. Through extensive experiments on fifteen publicly available datasets, we show that the TabTransformer outperforms the state-of-the-art deep learning methods for tabular data by at least 1.0% on mean AUC, and matches the performance of tree-based ensemble models. Furt","authors_text":"Ashish Khetan, Milan Cvitkovic, Xin Huang, Zohar Karnin","cross_cats":["cs.AI"],"headline":"TabTransformer applies self-attention to categorical feature embeddings to create contextual representations that raise prediction accuracy on tabular data.","license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-11T23:31:23Z","title":"TabTransformer: Tabular Data Modeling Using Contextual Embeddings"},"references":{"count":99,"internal_anchors":8,"resolved_work":99,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Proceedings of the ninth international conference on Information and knowledge management , pages=","work_id":"2a654019-2a19-4a40-8ad3-d71bd6cd4b0d","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Advances in neural information processing systems , pages=","work_id":"de2dd579-e4e4-4930-ab73-1cac1aa46d25","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Learning from labeled and unlabeled data with label propagation , author=. 2002 , publisher=","work_id":"af0992c3-e96a-451e-bc3e-e4a141a7dc29","year":2002},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Advances in neural information processing systems , pages=","work_id":"2679e2f3-f246-4196-aac3-0b15eed5727c","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Workshop on challenges in representation learning, ICML , volume=","work_id":"eb36ae7f-1e24-491c-aef6-39195c0a348c","year":null}],"snapshot_sha256":"287e376195552d6048286aa0be77469c57fb3405ac05d0c0912f26c7e9626284"},"source":{"id":"2012.06678","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-16T21:28:50.990411Z","id":"81d9d466-28f4-4742-8bc7-4d39225a178e","model_set":{"reader":"grok-4.3"},"one_line_summary":"TabTransformer uses Transformer self-attention to generate contextual embeddings from categorical features in tabular data, outperforming prior deep learning methods by at least 1% mean AUC and matching tree-based ensembles on 15 public datasets while showing robustness to missing and noisy features","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"TabTransformer applies self-attention to categorical feature embeddings to create contextual representations that raise prediction accuracy on tabular data.","strongest_claim":"Through extensive experiments on fifteen publicly available datasets, we show that the TabTransformer outperforms the state-of-the-art deep learning methods for tabular data by at least 1.0% on mean AUC, and matches the performance of tree-based ensemble models.","weakest_assumption":"The fifteen public datasets are representative of real-world tabular distributions and that baseline deep learning and tree methods were tuned to their best possible performance without hidden advantages for the proposed model."}},"verdict_id":"81d9d466-28f4-4742-8bc7-4d39225a178e"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:5ef60aa6fdf156230434f1ee8e92a08a69aef5377c590bcdacdb9d27c6fd0ea4","target":"record","created_at":"2026-05-17T23:38:46Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"14a98b2ef24421bfa593b9c24470b298a492a660bbbab01d932a8b622d357324","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2020-12-11T23:31:23Z","title_canon_sha256":"da775df861a4305faae84048f4463e76aecd6af3620932a75a861ab369a98ff6"},"schema_version":"1.0","source":{"id":"2012.06678","kind":"arxiv","version":1}},"canonical_sha256":"d6f8a8f862828579b3fa258a6002e6ed6ab6c0eac508138209d7c435e53f440a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d6f8a8f862828579b3fa258a6002e6ed6ab6c0eac508138209d7c435e53f440a","first_computed_at":"2026-05-17T23:38:46.506714Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:46.506714Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Gng8JZn3OYRs91zT2bO6cHnJuU9+FuaIWhOnpTknOKkMBhk3QSatB1DgVvkg++9iJs9kN1y/DEkvTYmwp0VICQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:46.507257Z","signed_message":"canonical_sha256_bytes"},"source_id":"2012.06678","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5ef60aa6fdf156230434f1ee8e92a08a69aef5377c590bcdacdb9d27c6fd0ea4","sha256:c5317c9a4914b6d354a6d91b622e29de49804d0730aa6edecf310564533ea145"],"state_sha256":"41a423462a20ea3fb44a8c223126c00593a82a023783a7e7d1575921d6f4bc2c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AUvhXSYSg5rsRxqbA+3iPsFWvaZ7SvSuZZTBraVOsyl9AUqPJyV/xhTkMhctHLGpEAybjiUOdLWXMGUzZCruCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T23:20:25.404056Z","bundle_sha256":"5f82969c9931c0e3042f57ff20ea1cbb5c8856b25f39a64cd527aa11f6fe4c79"}}