{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:WXM3UFDFJOQXKNATW2NS3T5Z23","short_pith_number":"pith:WXM3UFDF","schema_version":"1.0","canonical_sha256":"b5d9ba14654ba1753413b69b2dcfb9d6cdb8eb8b19d13cb98c699739ab0282ab","source":{"kind":"arxiv","id":"2109.03764","version":1},"attestation_state":"computed","paper":{"title":"Active Learning by Acquiring Contrastive Examples","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Giorgos Vernikos, Katerina Margatina, Lo\\\"ic Barrault, Nikolaos Aletras","submitted_at":"2021-09-08T16:40:18Z","abstract_excerpt":"Common acquisition functions for active learning use either uncertainty or diversity sampling, aiming to select difficult and diverse data points from the pool of unlabeled data, respectively. In this work, leveraging the best of both worlds, we propose an acquisition function that opts for selecting \\textit{contrastive examples}, i.e. data points that are similar in the model feature space and yet the model outputs maximally different predictive likelihoods. We compare our approach, CAL (Contrastive Active Learning), with a diverse set of acquisition functions in four natural language underst"},"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":"2109.03764","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-09-08T16:40:18Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"f0796aef052ae9f9292a6f17dce0c9f47d36f1ccc323abc8ea64ef40fde9611e","abstract_canon_sha256":"32c18154f488df334601b236882dfcc3b4cb22a618cb0395e67ef4f8e736df3e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:12:42.880573Z","signature_b64":"COq7+6q9DYxPgwymnrezEudGOaaBwO6wb64HA+nOrNcY16WhiL74Gk7LYJKN1EktCRcEAUS9b/mhH1EV34ZZCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b5d9ba14654ba1753413b69b2dcfb9d6cdb8eb8b19d13cb98c699739ab0282ab","last_reissued_at":"2026-07-05T03:12:42.880124Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:12:42.880124Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Active Learning by Acquiring Contrastive Examples","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Giorgos Vernikos, Katerina Margatina, Lo\\\"ic Barrault, Nikolaos Aletras","submitted_at":"2021-09-08T16:40:18Z","abstract_excerpt":"Common acquisition functions for active learning use either uncertainty or diversity sampling, aiming to select difficult and diverse data points from the pool of unlabeled data, respectively. In this work, leveraging the best of both worlds, we propose an acquisition function that opts for selecting \\textit{contrastive examples}, i.e. data points that are similar in the model feature space and yet the model outputs maximally different predictive likelihoods. We compare our approach, CAL (Contrastive Active Learning), with a diverse set of acquisition functions in four natural language underst"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.03764","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/2109.03764/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":"2109.03764","created_at":"2026-07-05T03:12:42.880186+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.03764v1","created_at":"2026-07-05T03:12:42.880186+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.03764","created_at":"2026-07-05T03:12:42.880186+00:00"},{"alias_kind":"pith_short_12","alias_value":"WXM3UFDFJOQX","created_at":"2026-07-05T03:12:42.880186+00:00"},{"alias_kind":"pith_short_16","alias_value":"WXM3UFDFJOQXKNAT","created_at":"2026-07-05T03:12:42.880186+00:00"},{"alias_kind":"pith_short_8","alias_value":"WXM3UFDF","created_at":"2026-07-05T03:12:42.880186+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08574","citing_title":"OrderDP: A Theoretically Guaranteed Lossless Dynamic Data Pruning Framework","ref_index":110,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23482","citing_title":"Multimodal Distribution Matching for Vision-Language Dataset Distillation","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12303","citing_title":"Labeled TrustSet Guided: Batch Active Learning with Reinforcement Learning","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23","json":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23.json","graph_json":"https://pith.science/api/pith-number/WXM3UFDFJOQXKNATW2NS3T5Z23/graph.json","events_json":"https://pith.science/api/pith-number/WXM3UFDFJOQXKNATW2NS3T5Z23/events.json","paper":"https://pith.science/paper/WXM3UFDF"},"agent_actions":{"view_html":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23","download_json":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23.json","view_paper":"https://pith.science/paper/WXM3UFDF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.03764&json=true","fetch_graph":"https://pith.science/api/pith-number/WXM3UFDFJOQXKNATW2NS3T5Z23/graph.json","fetch_events":"https://pith.science/api/pith-number/WXM3UFDFJOQXKNATW2NS3T5Z23/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23/action/storage_attestation","attest_author":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23/action/author_attestation","sign_citation":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23/action/citation_signature","submit_replication":"https://pith.science/pith/WXM3UFDFJOQXKNATW2NS3T5Z23/action/replication_record"}},"created_at":"2026-07-05T03:12:42.880186+00:00","updated_at":"2026-07-05T03:12:42.880186+00:00"}