{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:QFSDJM53LFN35MRTZPNZBDGW6D","short_pith_number":"pith:QFSDJM53","schema_version":"1.0","canonical_sha256":"816434b3bb595bbeb233cbdb908cd6f0e5d0821e7b6e3df7b7163d5bde3b6eff","source":{"kind":"arxiv","id":"2210.02748","version":1},"attestation_state":"computed","paper":{"title":"CLAD: A Contrastive Learning based Approach for Background Debiasing","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","eess.IV"],"primary_cat":"cs.CV","authors_text":"Harshitha Machiraju, Ke Wang, Michael Herzog, Oh-Hyeon Choung, Pascal Frossard","submitted_at":"2022-10-06T08:33:23Z","abstract_excerpt":"Convolutional neural networks (CNNs) have achieved superhuman performance in multiple vision tasks, especially image classification. However, unlike humans, CNNs leverage spurious features, such as background information to make decisions. This tendency creates different problems in terms of robustness or weak generalization performance. Through our work, we introduce a contrastive learning-based approach (CLAD) to mitigate the background bias in CNNs. CLAD encourages semantic focus on object foregrounds and penalizes learning features from irrelavant backgrounds. Our method also introduces an"},"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":"2210.02748","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-10-06T08:33:23Z","cross_cats_sorted":["cs.AI","cs.LG","eess.IV"],"title_canon_sha256":"1af3fdf50d8d977a3647b4186b03ae27b6bfdc1a82cb7801832604c583425408","abstract_canon_sha256":"69120116be6aa0dd6d2672a769feb4652e624f85c7e653fec53e406106812403"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:03:56.719750Z","signature_b64":"CBDp3LlugtWQm85C8zNwCKDYrXNLH716W0zmyUy9cgY8Ol0CDaeX9S299XJISnasr6FKcVcPYX1XO4igrGzEAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"816434b3bb595bbeb233cbdb908cd6f0e5d0821e7b6e3df7b7163d5bde3b6eff","last_reissued_at":"2026-07-05T05:03:56.719353Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:03:56.719353Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CLAD: A Contrastive Learning based Approach for Background Debiasing","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","eess.IV"],"primary_cat":"cs.CV","authors_text":"Harshitha Machiraju, Ke Wang, Michael Herzog, Oh-Hyeon Choung, Pascal Frossard","submitted_at":"2022-10-06T08:33:23Z","abstract_excerpt":"Convolutional neural networks (CNNs) have achieved superhuman performance in multiple vision tasks, especially image classification. However, unlike humans, CNNs leverage spurious features, such as background information to make decisions. This tendency creates different problems in terms of robustness or weak generalization performance. Through our work, we introduce a contrastive learning-based approach (CLAD) to mitigate the background bias in CNNs. CLAD encourages semantic focus on object foregrounds and penalizes learning features from irrelavant backgrounds. Our method also introduces an"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.02748","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/2210.02748/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":"2210.02748","created_at":"2026-07-05T05:03:56.719413+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.02748v1","created_at":"2026-07-05T05:03:56.719413+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.02748","created_at":"2026-07-05T05:03:56.719413+00:00"},{"alias_kind":"pith_short_12","alias_value":"QFSDJM53LFN3","created_at":"2026-07-05T05:03:56.719413+00:00"},{"alias_kind":"pith_short_16","alias_value":"QFSDJM53LFN35MRT","created_at":"2026-07-05T05:03:56.719413+00:00"},{"alias_kind":"pith_short_8","alias_value":"QFSDJM53","created_at":"2026-07-05T05:03:56.719413+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09495","citing_title":"ContextShift: A Controlled Benchmark for Context Dependence in Object Detection","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D","json":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D.json","graph_json":"https://pith.science/api/pith-number/QFSDJM53LFN35MRTZPNZBDGW6D/graph.json","events_json":"https://pith.science/api/pith-number/QFSDJM53LFN35MRTZPNZBDGW6D/events.json","paper":"https://pith.science/paper/QFSDJM53"},"agent_actions":{"view_html":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D","download_json":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D.json","view_paper":"https://pith.science/paper/QFSDJM53","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.02748&json=true","fetch_graph":"https://pith.science/api/pith-number/QFSDJM53LFN35MRTZPNZBDGW6D/graph.json","fetch_events":"https://pith.science/api/pith-number/QFSDJM53LFN35MRTZPNZBDGW6D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D/action/storage_attestation","attest_author":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D/action/author_attestation","sign_citation":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D/action/citation_signature","submit_replication":"https://pith.science/pith/QFSDJM53LFN35MRTZPNZBDGW6D/action/replication_record"}},"created_at":"2026-07-05T05:03:56.719413+00:00","updated_at":"2026-07-05T05:03:56.719413+00:00"}