{"id":"W3171439159","doi":"10.1109/access.2021.3086321","title":"Substituting Convolutions for Neural Network Compression","year":2021,"lang":"en","type":"article","venue":"IEEE Access","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Staatssekretariat für Bildung, Forschung und Innovation; Engineering and Physical Sciences Research Council; European Commission","keywords":"Computer science; Pointwise; Leverage (statistics); Hyperparameter; Convolutional neural network; Artificial neural network; Network architecture; Data compression ratio; Simple (philosophy); Pareto principle; Computer engineering; Artificial intelligence; Machine learning; Mathematical optimization; Image compression; Mathematics; Computer network","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":[0.00007501237,0.0001084859,0.000132467,0.0000228605,0.0005139609,0.0002212608,0.0009778364,0.00004260322,0.000006627582],"category_scores_gemma":[0.00003281012,0.0001094407,0.00006685287,0.0006511169,0.00004205866,0.000795753,0.0002964754,0.0001177213,0.00001194424],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002670969,"about_ca_system_score_gemma":0.00005538764,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003974604,"about_ca_topic_score_gemma":0.00001934194,"domain_scores_codex":[0.9988708,0.00003786835,0.0002056102,0.0004042206,0.0001245977,0.0003569419],"domain_scores_gemma":[0.9987366,0.0003355374,0.0000994227,0.0005730808,0.0001693686,0.00008597865],"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":[0.000008764378,0.00008530793,0.002316329,0.0000258064,0.00001948205,0.0000210839,0.00006332796,0.7484115,0.006595126,0.1798035,0.02997766,0.03267206],"study_design_scores_gemma":[0.0004786486,0.00001618351,0.003402757,0.00004440817,0.00001321935,0.00004062237,0.000004853741,0.9007111,0.01524281,0.03919825,0.04053021,0.0003169068],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01871823,0.0003826448,0.9772847,0.001323192,0.001207321,0.0002683346,0.000004863578,0.0002495545,0.0005611513],"genre_scores_gemma":[0.9259151,0.00002239158,0.07204781,0.001042212,0.0005891734,0.0001577969,0.0000151991,0.00001186612,0.0001984345],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9071969,"threshold_uncertainty_score":0.4462863,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06417957861793859,"score_gpt":0.3479631749883805,"score_spread":0.2837835963704419,"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."}}