{"id":"W4385570917","doi":"10.18653/v1/2023.findings-acl.867","title":"Model Interpretability and Rationale Extraction by Input Mask Optimization","year":2023,"lang":"en","type":"article","venue":"","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Interpretability; Computer science; Artificial intelligence; Machine learning; Regularization (linguistics); Artificial neural network; Masking (illustration); Modalities; Task (project management); Pattern recognition (psychology)","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007567699,0.001479586,0.001291011,0.002684282,0.0009821085,0.004648621,0.002223269,0.001547512,0.009984786],"category_scores_gemma":[0.04490228,0.001185467,0.003537062,0.001835618,0.001913931,0.005304873,0.005158263,0.00287494,0.002542082],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001712751,"about_ca_system_score_gemma":0.002937036,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004329364,"about_ca_topic_score_gemma":0.004718797,"domain_scores_codex":[0.9925655,0.003928904,0.0006606358,0.001064899,0.001446975,0.0003330642],"domain_scores_gemma":[0.9745059,0.01860813,0.0008212326,0.00401744,0.001827295,0.0002200779],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001121904,0.0002702634,0.005469142,0.001122528,0.0004636574,0.0009238762,0.00189791,0.1447379,0.01671338,0.2500395,0.0231583,0.5540817],"study_design_scores_gemma":[0.00006998765,0.00005288369,0.0006807515,0.0001832071,0.0002053557,0.0001429267,0.0002618117,0.7060032,0.01260481,0.2680618,0.01168477,0.00004851395],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01756279,0.0004877503,0.9683921,0.001525005,0.0000778113,0.0003080585,0.001623257,0.003080146,0.00694313],"genre_scores_gemma":[0.2601433,0.000304495,0.7313864,0.0002255102,0.00005608177,0.0003837839,0.003911871,0.001068804,0.002519745],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.009984786,"threshold_uncertainty_score":0.04002231,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03587866272525021,"score_gpt":0.2899735662333752,"score_spread":0.254094903508125,"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."}}