{"id":"W1877062207","doi":"","title":"Scaling up Natural Gradient by Sparsely Factorizing the Inverse Fisher Matrix","year":2015,"lang":"en","type":"article","venue":"","topic":"Face and Expression Recognition","field":"Computer Science","cited_by":47,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Stochastic gradient descent; Gradient descent; Covariance matrix; Algorithm; Scaling; Fisher information; Gaussian; Mathematics; Matrix (chemical analysis); Applied mathematics; Computer science; Inverse; Mathematical optimization; Gradient method; Artificial intelligence; Statistics; Artificial neural 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":[0.0002078667,0.00009109127,0.0000724704,0.00003056231,0.0001090221,0.0002024065,0.0004412864,0.00003900264,0.00002977404],"category_scores_gemma":[0.0000437344,0.00005299273,0.00004243735,0.0001635722,0.00002189611,0.0005231379,0.0001669526,0.0001343346,0.0002975637],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003188808,"about_ca_system_score_gemma":0.00002543029,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000122808,"about_ca_topic_score_gemma":0.00001811258,"domain_scores_codex":[0.9991649,0.00004844344,0.0001211543,0.0002029917,0.0002567005,0.0002058051],"domain_scores_gemma":[0.9994841,0.00005358058,0.00004184169,0.0002482289,0.00006451948,0.0001077836],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000006580466,0.00002381349,0.0003837737,0.000003782696,0.00001021712,0.000003815243,0.002161196,0.00004979658,0.002998431,0.00135372,0.9761671,0.01683781],"study_design_scores_gemma":[0.001523081,0.00008897713,0.0005079051,0.00007791363,0.00001704583,0.00003723226,0.002916756,0.3245049,0.05098216,0.005664845,0.6129451,0.0007340278],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7649762,0.0005814467,0.201921,0.0096791,0.008716708,0.0004127247,0.000005679655,0.0006774745,0.01302965],"genre_scores_gemma":[0.9851391,0.00001139154,0.007197693,0.001479324,0.00008118643,0.000007914542,0.000006848294,0.000006772305,0.006069746],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3632219,"threshold_uncertainty_score":0.3824679,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.068452524059194,"score_gpt":0.2633277899671421,"score_spread":0.1948752659079481,"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."}}