{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003290521,0.00007612021,0.00006884326,0.0000803743,0.0001180122,0.0001660354,0.0002137854,0.00004598756,0.00004844866],"category_scores_gemma":[0.00009735385,0.00007455723,0.0000196994,0.0004021152,0.0000347524,0.00125948,0.0001405158,0.00006499171,0.00009604562],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003995542,"about_ca_system_score_gemma":0.00003445494,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003814803,"about_ca_topic_score_gemma":0.00002241216,"domain_scores_codex":[0.9991457,0.00003704623,0.0001809043,0.0003146844,0.0001648527,0.0001567392],"domain_scores_gemma":[0.9994648,0.0001023108,0.00003894997,0.0002580349,0.00007918701,0.00005678548],"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":[0.000006810744,0.00004965075,0.0003192271,0.00001108205,0.00000549951,0.000001575512,0.001327777,0.9081273,0.007315967,0.05579274,0.006116399,0.02092596],"study_design_scores_gemma":[0.00002699722,0.0000161102,0.00003868095,0.00000270508,0.00000113715,0.000001847955,0.0001018199,0.9721113,0.008845802,0.01856639,0.0002048934,0.0000823083],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02447769,0.00001706988,0.9699167,0.001879979,0.0001083669,0.0001538873,0.000001873874,0.0003378852,0.003106547],"genre_scores_gemma":[0.9125348,0.00004581884,0.08460701,0.0003016147,0.00001615951,0.00003188869,0.0000138638,0.000006541002,0.002442347],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8880571,"threshold_uncertainty_score":0.3040355,"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."}}