{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:5O2ZB33RXHOWPSIKMO4FM6PMD7","short_pith_number":"pith:5O2ZB33R","schema_version":"1.0","canonical_sha256":"ebb590ef71b9dd67c90a63b85679ec1fd86ab9f0839f71f722e953363b4bfca4","source":{"kind":"arxiv","id":"2102.07437","version":3},"attestation_state":"computed","paper":{"title":"Data Quality Matters For Adversarial Training: An Empirical Study","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chengyu Dong, Jingbo Shang, Liyuan Liu","submitted_at":"2021-02-15T10:17:24Z","abstract_excerpt":"Multiple intriguing problems are hovering in adversarial training, including robust overfitting, robustness overestimation, and robustness-accuracy trade-off. These problems pose great challenges to both reliable evaluation and practical deployment. Here, we empirically show that these problems share one common cause -- low-quality samples in the dataset. Specifically, we first propose a strategy to measure the data quality based on the learning behaviors of the data during adversarial training and find that low-quality data may not be useful and even detrimental to the adversarial robustness."},"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":"2102.07437","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-02-15T10:17:24Z","cross_cats_sorted":[],"title_canon_sha256":"c250e174486b3a70c90d09dae1852157e66a5ffe14b1a692bc09dcebdc2566a4","abstract_canon_sha256":"1e21314b0a4f8fe8450dbb6f66ca25f5bb33bbb9bd563962cd1e470ab08024c2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:20:34.033640Z","signature_b64":"bXbeJ0n28Z6cNKrZe+PxNocIaJOEDt3vvsbOI7ZUc7KSv4Tn4Y+Fxl84GckJBgsUBhRsPGlDS25PW4/QL4+zCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ebb590ef71b9dd67c90a63b85679ec1fd86ab9f0839f71f722e953363b4bfca4","last_reissued_at":"2026-07-05T03:20:34.033241Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:20:34.033241Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Data Quality Matters For Adversarial Training: An Empirical Study","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chengyu Dong, Jingbo Shang, Liyuan Liu","submitted_at":"2021-02-15T10:17:24Z","abstract_excerpt":"Multiple intriguing problems are hovering in adversarial training, including robust overfitting, robustness overestimation, and robustness-accuracy trade-off. These problems pose great challenges to both reliable evaluation and practical deployment. Here, we empirically show that these problems share one common cause -- low-quality samples in the dataset. Specifically, we first propose a strategy to measure the data quality based on the learning behaviors of the data during adversarial training and find that low-quality data may not be useful and even detrimental to the adversarial robustness."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.07437","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/2102.07437/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":"2102.07437","created_at":"2026-07-05T03:20:34.033297+00:00"},{"alias_kind":"arxiv_version","alias_value":"2102.07437v3","created_at":"2026-07-05T03:20:34.033297+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.07437","created_at":"2026-07-05T03:20:34.033297+00:00"},{"alias_kind":"pith_short_12","alias_value":"5O2ZB33RXHOW","created_at":"2026-07-05T03:20:34.033297+00:00"},{"alias_kind":"pith_short_16","alias_value":"5O2ZB33RXHOWPSIK","created_at":"2026-07-05T03:20:34.033297+00:00"},{"alias_kind":"pith_short_8","alias_value":"5O2ZB33R","created_at":"2026-07-05T03:20:34.033297+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/5O2ZB33RXHOWPSIKMO4FM6PMD7","json":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7.json","graph_json":"https://pith.science/api/pith-number/5O2ZB33RXHOWPSIKMO4FM6PMD7/graph.json","events_json":"https://pith.science/api/pith-number/5O2ZB33RXHOWPSIKMO4FM6PMD7/events.json","paper":"https://pith.science/paper/5O2ZB33R"},"agent_actions":{"view_html":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7","download_json":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7.json","view_paper":"https://pith.science/paper/5O2ZB33R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2102.07437&json=true","fetch_graph":"https://pith.science/api/pith-number/5O2ZB33RXHOWPSIKMO4FM6PMD7/graph.json","fetch_events":"https://pith.science/api/pith-number/5O2ZB33RXHOWPSIKMO4FM6PMD7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7/action/storage_attestation","attest_author":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7/action/author_attestation","sign_citation":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7/action/citation_signature","submit_replication":"https://pith.science/pith/5O2ZB33RXHOWPSIKMO4FM6PMD7/action/replication_record"}},"created_at":"2026-07-05T03:20:34.033297+00:00","updated_at":"2026-07-05T03:20:34.033297+00:00"}