{"id":"W4385571067","doi":"10.18653/v1/2023.acl-long.483","title":"Efficient Shapley Values Estimation by Amortization for Text Classification","year":2023,"lang":"en","type":"article","venue":"","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Samsung; Institute for Catastrophic Loss Reduction; Cisco Systems","keywords":"Shapley value; Interpretability; Computer science; Stability (learning theory); Amortized analysis; Feature (linguistics); Theoretical computer science; Algorithm; Artificial intelligence; Machine learning; Mathematics; Data structure; Mathematical economics; Game theory","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":[],"consensus_categories":[],"category_scores_codex":[0.0005348001,0.0001048521,0.00009281796,0.0001655786,0.0002488156,0.0002183975,0.0004472619,0.00005772389,0.00002233464],"category_scores_gemma":[0.0002926106,0.0001003457,0.00004587465,0.00102537,0.00003243738,0.0003049175,0.0000763568,0.00004053455,0.0007556063],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000747386,"about_ca_system_score_gemma":0.00004122454,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003064437,"about_ca_topic_score_gemma":0.000006885515,"domain_scores_codex":[0.9987119,0.00003629099,0.0002958062,0.0003957814,0.0002736572,0.000286524],"domain_scores_gemma":[0.99905,0.0002030753,0.0001008813,0.0003865988,0.0001954247,0.00006406472],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000008660464,0.0001226161,0.0001666586,0.00003557561,0.00001055334,9.10017e-7,0.001532697,0.1679554,0.02988374,0.5852525,0.05959957,0.1554312],"study_design_scores_gemma":[0.00005322333,0.00004560079,0.0006640841,0.000007141141,0.000003060148,5.292262e-7,0.0001698816,0.950043,0.03420107,0.01352782,0.001163167,0.0001213773],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02847332,0.00001397756,0.9661747,0.002455063,0.0002791849,0.0004744231,0.000004058887,0.0007789958,0.00134632],"genre_scores_gemma":[0.9565353,0.000007115782,0.04074632,0.0002114525,0.00003986582,0.0001698218,0.00008210639,0.00001399049,0.002193989],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.928062,"threshold_uncertainty_score":0.9712042,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0499369776387494,"score_gpt":0.3194781850968632,"score_spread":0.2695412074581138,"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."}}