{"id":"W3160314846","doi":"10.1109/cvpr52688.2022.01434","title":"When Does Contrastive Visual Representation Learning Work?","year":2022,"lang":"en","type":"article","venue":"2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":98,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Resnick Sustainability Institute for Science, Energy and Sustainability, California Institute of Technology; McGill University; California Institute of Technology; Danmarks Grundforskningsfond; National Science Foundation","keywords":"Computer science; Artificial intelligence; Supervised learning; Machine learning; Granularity; Task (project management); Representation (politics); Semi-supervised learning; Feature learning; Field (mathematics); Unsupervised learning; Labeled data; Domain (mathematical analysis); Natural language processing; Pattern recognition (psychology); Artificial neural network; Mathematics","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006136494,0.0003309084,0.0003708188,0.000369261,0.000940933,0.0008023606,0.0004910263,0.00008400816,0.002220969],"category_scores_gemma":[0.00003419804,0.0003011228,0.0001219161,0.0004406111,0.00007674372,0.0006375369,0.0004563309,0.0008530382,0.0002298118],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007371219,"about_ca_system_score_gemma":0.00005654311,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003447777,"about_ca_topic_score_gemma":0.000007536409,"domain_scores_codex":[0.9963484,0.0009851431,0.000496254,0.0009883819,0.0007734967,0.0004083389],"domain_scores_gemma":[0.9985141,0.0003983947,0.0003568176,0.0002948584,0.0002118884,0.0002239537],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001208953,0.000189743,0.001419159,0.00001636942,0.00004032599,0.00006774455,0.003048439,0.0007301758,0.0005777424,0.0004743778,0.002195917,0.9911191],"study_design_scores_gemma":[0.00396093,0.002799612,0.01635601,0.0003195748,0.00003904089,0.0001275817,0.002157554,0.9479278,0.001025658,0.006154426,0.01773947,0.001392305],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1580667,0.00003657594,0.8347183,0.002462072,0.001959922,0.0004492956,0.00002034069,0.0004158385,0.001870914],"genre_scores_gemma":[0.991923,0.00007736425,0.003395844,0.003231016,0.0002256013,0.00009247991,0.0001808163,0.00002992169,0.0008439289],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9897268,"threshold_uncertainty_score":0.9999441,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0420139713652861,"score_gpt":0.290573691307519,"score_spread":0.2485597199422329,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}