{"id":"W3189787503","doi":"10.1109/jsait.2021.3103494","title":"Compressing Gradients by Exploiting Temporal Correlation in Momentum-SGD","year":2021,"lang":"en","type":"article","venue":"IEEE Journal on Selected Areas in Information Theory","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Huawei Technologies","keywords":"Bottleneck; Computer science; Computation; Rate of convergence; Convergence (economics); Algorithm; Momentum (technical analysis); Mathematical optimization; Information bottleneck method; Compression (physics); Exploit; Norm (philosophy); Mathematics; Artificial intelligence; Telecommunications","routes":{"ca_aff":true,"ca_fund":true,"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":[0.0009373337,0.0001723078,0.0002005542,0.0007713391,0.000174259,0.0004002289,0.0003956135,0.0001008065,0.00002429421],"category_scores_gemma":[0.0006619653,0.0001772786,0.00004371789,0.001487435,0.00002869854,0.003263063,0.00004965105,0.0005661909,0.00001958586],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004048629,"about_ca_system_score_gemma":0.0001567514,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003467146,"about_ca_topic_score_gemma":0.000001240588,"domain_scores_codex":[0.997796,0.0003451381,0.0008812094,0.0001646015,0.000492418,0.0003206313],"domain_scores_gemma":[0.9983969,0.0003316251,0.0005318727,0.0002213402,0.00041835,0.00009991258],"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":[0.0004456339,0.001328279,0.02573592,0.0001087951,0.000116737,0.0002052999,0.0288048,0.4186393,0.003383747,0.360283,0.01683189,0.1441166],"study_design_scores_gemma":[0.003490507,0.0002735909,0.00522556,0.001097551,0.00000990922,0.0004862486,0.001114198,0.895459,0.02181629,0.06906649,0.001188775,0.0007718438],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0589061,0.00004642472,0.937984,0.0001741607,0.0006981412,0.0001647017,0.000003606049,0.0001473417,0.001875489],"genre_scores_gemma":[0.9829719,0.00003171093,0.01615382,0.0006655114,0.00003741644,0.00002179992,0.00005215062,0.00001081859,0.00005485014],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9240658,"threshold_uncertainty_score":0.722921,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.009529290614629546,"score_gpt":0.2318131972925966,"score_spread":0.222283906677967,"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."}}