{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001723052,0.0001509446,0.0001262502,0.00007305019,0.0003492484,0.000160368,0.0005666767,0.00006827367,0.0001124815],"category_scores_gemma":[0.0000285651,0.000161728,0.00006012306,0.0003945234,0.0001755477,0.001202308,0.0001623544,0.00009501562,0.0002220348],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007897158,"about_ca_system_score_gemma":0.00004337008,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003771866,"about_ca_topic_score_gemma":0.00002678633,"domain_scores_codex":[0.9988779,0.000104705,0.0001125452,0.0005423271,0.00006979505,0.0002927447],"domain_scores_gemma":[0.9991932,0.00004939995,0.00008219219,0.0004114685,0.0001477179,0.0001160549],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001490773,0.0003680634,0.002052906,0.00001623139,0.0003452767,0.0002997606,0.003893491,0.06360668,0.3882653,0.1449827,0.374423,0.02159761],"study_design_scores_gemma":[0.0003242707,0.0001145569,0.00007546095,0.000005209197,0.00001191142,0.000002309496,0.0001420611,0.9638926,0.0275776,0.0004242263,0.00720639,0.0002233554],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1007381,0.00001384878,0.8943404,0.0001165396,0.000419537,0.00006751294,0.00000432647,0.00009686851,0.004202847],"genre_scores_gemma":[0.9928722,0.00002072208,0.00524053,0.0002351116,0.0002928599,3.263768e-7,0.000006299227,0.000009071666,0.001322855],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.900286,"threshold_uncertainty_score":0.6595078,"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."}}