{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:U62JQZGKKKL6TOTAX5Q5FLF23C","short_pith_number":"pith:U62JQZGK","schema_version":"1.0","canonical_sha256":"a7b49864ca5297e9ba60bf61d2acbad88e3e6a2811d6d25ad642c8145e13c1ba","source":{"kind":"arxiv","id":"2401.01764","version":1},"attestation_state":"computed","paper":{"title":"Understanding the Detrimental Class-level Effects of Data Augmentation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Andrew Gordon Wilson, Diane Bouchacourt, Hamed Firooz, Mark Ibrahim, Polina Kirichenko, Ramakrishna Vedantam, Randall Balestriero","submitted_at":"2023-12-07T18:37:43Z","abstract_excerpt":"Data augmentation (DA) encodes invariance and provides implicit regularization critical to a model's performance in image classification tasks. However, while DA improves average accuracy, recent studies have shown that its impact can be highly class dependent: achieving optimal average accuracy comes at the cost of significantly hurting individual class accuracy by as much as 20% on ImageNet. There has been little progress in resolving class-level accuracy drops due to a limited understanding of these effects. In this work, we present a framework for understanding how DA interacts with class-"},"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":"2401.01764","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-12-07T18:37:43Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"9fbffa653f0b2280eb76536330f1b2e7755af1c3402022d508a16b8bc830a203","abstract_canon_sha256":"f0bd196a50ee630fbd9586990a43fe15b7c4c9bcaf4ebc1053a047cc6c3f5546"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:29:56.941647Z","signature_b64":"j9aB8znJEK0Xq34W2uurueqxEJVQZDt91aLmei4NbwcrEymx9AsAcsDZbBDqVdmRGRWNRYTpWFuxDpQF7OawBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a7b49864ca5297e9ba60bf61d2acbad88e3e6a2811d6d25ad642c8145e13c1ba","last_reissued_at":"2026-07-05T07:29:56.941223Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:29:56.941223Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding the Detrimental Class-level Effects of Data Augmentation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Andrew Gordon Wilson, Diane Bouchacourt, Hamed Firooz, Mark Ibrahim, Polina Kirichenko, Ramakrishna Vedantam, Randall Balestriero","submitted_at":"2023-12-07T18:37:43Z","abstract_excerpt":"Data augmentation (DA) encodes invariance and provides implicit regularization critical to a model's performance in image classification tasks. However, while DA improves average accuracy, recent studies have shown that its impact can be highly class dependent: achieving optimal average accuracy comes at the cost of significantly hurting individual class accuracy by as much as 20% on ImageNet. There has been little progress in resolving class-level accuracy drops due to a limited understanding of these effects. In this work, we present a framework for understanding how DA interacts with class-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.01764","kind":"arxiv","version":1},"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/2401.01764/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":"2401.01764","created_at":"2026-07-05T07:29:56.941277+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.01764v1","created_at":"2026-07-05T07:29:56.941277+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.01764","created_at":"2026-07-05T07:29:56.941277+00:00"},{"alias_kind":"pith_short_12","alias_value":"U62JQZGKKKL6","created_at":"2026-07-05T07:29:56.941277+00:00"},{"alias_kind":"pith_short_16","alias_value":"U62JQZGKKKL6TOTA","created_at":"2026-07-05T07:29:56.941277+00:00"},{"alias_kind":"pith_short_8","alias_value":"U62JQZGK","created_at":"2026-07-05T07:29:56.941277+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.06831","citing_title":"No Location Left Behind: Measuring and Improving the Fairness of Implicit Representations for Earth Data","ref_index":24,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C","json":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C.json","graph_json":"https://pith.science/api/pith-number/U62JQZGKKKL6TOTAX5Q5FLF23C/graph.json","events_json":"https://pith.science/api/pith-number/U62JQZGKKKL6TOTAX5Q5FLF23C/events.json","paper":"https://pith.science/paper/U62JQZGK"},"agent_actions":{"view_html":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C","download_json":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C.json","view_paper":"https://pith.science/paper/U62JQZGK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.01764&json=true","fetch_graph":"https://pith.science/api/pith-number/U62JQZGKKKL6TOTAX5Q5FLF23C/graph.json","fetch_events":"https://pith.science/api/pith-number/U62JQZGKKKL6TOTAX5Q5FLF23C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C/action/storage_attestation","attest_author":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C/action/author_attestation","sign_citation":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C/action/citation_signature","submit_replication":"https://pith.science/pith/U62JQZGKKKL6TOTAX5Q5FLF23C/action/replication_record"}},"created_at":"2026-07-05T07:29:56.941277+00:00","updated_at":"2026-07-05T07:29:56.941277+00:00"}