{"id":"W4385768183","doi":"10.24963/ijcai.2023/108","title":"IMF: Integrating Matched Features Using Attentive Logit in Knowledge Distillation","year":2023,"lang":"en","type":"article","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"Institute for Information and Communications Technology Promotion; Ministry of Science and ICT, South Korea; Sungkyunkwan University","keywords":"Softmax function; Computer science; Margin (machine learning); Distillation; Inference; Feature (linguistics); Machine learning; Artificial intelligence; Logit; Matching (statistics); Representation (politics); Pattern recognition (psychology); Artificial neural network; Mathematics; Statistics","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":[0.001302815,0.001437982,0.001091392,0.001271722,0.0005800746,0.001290543,0.002556315,0.001429191,0.004740441],"category_scores_gemma":[0.003930549,0.0004854653,0.0009992969,0.001200025,0.0008175027,0.004494495,0.00273584,0.002423676,0.001541442],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008013632,"about_ca_system_score_gemma":0.001543456,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006718997,"about_ca_topic_score_gemma":0.009081081,"domain_scores_codex":[0.9993445,0.0001256351,0.00003579313,0.000256161,0.0001351021,0.0001028323],"domain_scores_gemma":[0.999289,0.0003211936,0.00006147956,0.0001629132,0.0001150292,0.00005027289],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003941999,0.0005006597,0.004188749,0.0002131851,0.0001608516,0.0002553992,0.0002131147,0.17056,0.01078094,0.01357752,0.008177473,0.7909778],"study_design_scores_gemma":[0.00004133427,0.0001192844,0.000597518,0.00002441181,0.00005101799,0.00007567633,0.0000364532,0.9721716,0.006667814,0.01700028,0.003187173,0.00002753269],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05046308,0.0008942062,0.9359443,0.0005200873,0.0001085244,0.0001347496,0.0004987619,0.008387898,0.003048352],"genre_scores_gemma":[0.7132038,0.00037145,0.2757056,0.0006722589,0.000105052,0.0002057527,0.001837423,0.000292867,0.007605652],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006718997,"threshold_uncertainty_score":0.01585835,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04652173233494302,"score_gpt":0.333634815200821,"score_spread":0.287113082865878,"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."}}