{"id":"W2896125160","doi":"10.48550/arxiv.1807.08024","title":"Explaining Image Classifiers by Counterfactual Generation","year":2018,"lang":"en","type":"article","venue":"arXiv (Cornell University)","topic":"Generative Adversarial Networks and Image Synthesis","field":"Computer Science","cited_by":89,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Classifier (UML); Artificial intelligence; Computer science; Pattern recognition (psychology); Counterfactual thinking; Generative grammar; Generative model; Image (mathematics); Ask price; Contextual image classification; Computer vision; Machine learning","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001589728,0.0006483936,0.0005348471,0.0006962788,0.0003159582,0.001084194,0.001103324,0.001305652,0.003819357],"category_scores_gemma":[0.008480346,0.0005348858,0.0007889693,0.0004338746,0.001703446,0.001763254,0.001190574,0.001428986,0.0004366015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009139795,"about_ca_system_score_gemma":0.000409604,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002195799,"about_ca_topic_score_gemma":0.00234682,"domain_scores_codex":[0.9995601,0.0001895862,0.00001459981,0.0001077639,0.00008047307,0.0000474888],"domain_scores_gemma":[0.9965629,0.002582007,0.0002651763,0.000365582,0.0001519406,0.00007243009],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001751739,0.0000539207,0.002531779,0.0001500479,0.00007934648,0.0003275752,0.000412109,0.7113571,0.005909438,0.2086969,0.00415863,0.06614795],"study_design_scores_gemma":[0.00001241988,0.00001288012,0.000202749,0.000009776827,0.00000899095,0.00005306748,0.00001166825,0.9213084,0.0008608804,0.07681052,0.000701229,0.000007504149],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04346533,0.000357877,0.9501382,0.001303591,0.00007908112,0.00006167848,0.0001429807,0.0005453816,0.003906026],"genre_scores_gemma":[0.8887331,0.0003206457,0.1064155,0.0004593094,0.0001357201,0.0001434306,0.000229358,0.0001564963,0.003406392],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003819357,"threshold_uncertainty_score":0.01277697,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0629509378043967,"score_gpt":0.1820183008027113,"score_spread":0.1190673629983146,"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."}}