{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:NE3TZ5M2EWNNH4NHERQKBAJQP7","short_pith_number":"pith:NE3TZ5M2","schema_version":"1.0","canonical_sha256":"69373cf59a259ad3f1a72460a081307ffe946dd5e093053353875b1fb33d591b","source":{"kind":"arxiv","id":"2302.06015","version":3},"attestation_state":"computed","paper":{"title":"A Theoretical Understanding of Shallow Vision Transformers: Learning, Generalization, and Sample Complexity","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Hongkang Li, Meng Wang, Pin-Yu Chen, Sijia Liu","submitted_at":"2023-02-12T22:12:35Z","abstract_excerpt":"Vision Transformers (ViTs) with self-attention modules have recently achieved great empirical success in many vision tasks. Due to non-convex interactions across layers, however, theoretical learning and generalization analysis is mostly elusive. Based on a data model characterizing both label-relevant and label-irrelevant tokens, this paper provides the first theoretical analysis of training a shallow ViT, i.e., one self-attention layer followed by a two-layer perceptron, for a classification task. We characterize the sample complexity to achieve a zero generalization error. Our sample comple"},"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":"2302.06015","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-02-12T22:12:35Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"641d8ea5a58affadeea6f05b0375bf230a4b22d0bd7ebd0d67fd6cd2e9aa8a5e","abstract_canon_sha256":"267eeb5790528225dfc38b325a67ea8c0fc0a90417a48418cd29b3ba3509e41d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:12:16.055183Z","signature_b64":"0i0q92eIOPah/9iSwWsF6wgQ1eeDhWfE/s5kU7lNXnGAz4+F3D6X8yMFDYeF1A63gs8qLp9z0wuFhI2aQTnJAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"69373cf59a259ad3f1a72460a081307ffe946dd5e093053353875b1fb33d591b","last_reissued_at":"2026-07-05T07:12:16.054694Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:12:16.054694Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Theoretical Understanding of Shallow Vision Transformers: Learning, Generalization, and Sample Complexity","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Hongkang Li, Meng Wang, Pin-Yu Chen, Sijia Liu","submitted_at":"2023-02-12T22:12:35Z","abstract_excerpt":"Vision Transformers (ViTs) with self-attention modules have recently achieved great empirical success in many vision tasks. Due to non-convex interactions across layers, however, theoretical learning and generalization analysis is mostly elusive. Based on a data model characterizing both label-relevant and label-irrelevant tokens, this paper provides the first theoretical analysis of training a shallow ViT, i.e., one self-attention layer followed by a two-layer perceptron, for a classification task. We characterize the sample complexity to achieve a zero generalization error. Our sample comple"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.06015","kind":"arxiv","version":3},"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/2302.06015/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":"2302.06015","created_at":"2026-07-05T07:12:16.054753+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.06015v3","created_at":"2026-07-05T07:12:16.054753+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.06015","created_at":"2026-07-05T07:12:16.054753+00:00"},{"alias_kind":"pith_short_12","alias_value":"NE3TZ5M2EWNN","created_at":"2026-07-05T07:12:16.054753+00:00"},{"alias_kind":"pith_short_16","alias_value":"NE3TZ5M2EWNNH4NH","created_at":"2026-07-05T07:12:16.054753+00:00"},{"alias_kind":"pith_short_8","alias_value":"NE3TZ5M2","created_at":"2026-07-05T07:12:16.054753+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.19724","citing_title":"Benign Overfitting in Adversarial Training for Vision Transformers","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06440","citing_title":"Visual prompting reimagined: The power of the Activation Prompts","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21056","citing_title":"Radiomics-Guided Vision Transformers for Survival Analysis","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7","json":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7.json","graph_json":"https://pith.science/api/pith-number/NE3TZ5M2EWNNH4NHERQKBAJQP7/graph.json","events_json":"https://pith.science/api/pith-number/NE3TZ5M2EWNNH4NHERQKBAJQP7/events.json","paper":"https://pith.science/paper/NE3TZ5M2"},"agent_actions":{"view_html":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7","download_json":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7.json","view_paper":"https://pith.science/paper/NE3TZ5M2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.06015&json=true","fetch_graph":"https://pith.science/api/pith-number/NE3TZ5M2EWNNH4NHERQKBAJQP7/graph.json","fetch_events":"https://pith.science/api/pith-number/NE3TZ5M2EWNNH4NHERQKBAJQP7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7/action/storage_attestation","attest_author":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7/action/author_attestation","sign_citation":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7/action/citation_signature","submit_replication":"https://pith.science/pith/NE3TZ5M2EWNNH4NHERQKBAJQP7/action/replication_record"}},"created_at":"2026-07-05T07:12:16.054753+00:00","updated_at":"2026-07-05T07:12:16.054753+00:00"}