{"id":"W3196726140","doi":"10.1109/isit45174.2021.9518254","title":"Differentially Quantized Gradient Descent","year":2021,"lang":"en","type":"article","venue":"","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Russian Academy of Sciences; Natural Sciences and Engineering Research Council of Canada; Princeton University; Jet Propulsion Laboratory; Division of Mathematical Sciences; Moscow Institute of Physics and Technology; King Abdullah University of Science and Technology; University of Ottawa; Indian Institute of Science; University of Tehran; National Aeronautics and Space Administration; California Institute of Technology; Amgen; National Science Foundation","keywords":"Gradient descent; Stochastic gradient descent; Quantization (signal processing); Convergence (economics); Descent (aeronautics); Computer science; Algorithm; Mathematics; Discrete mathematics; Theoretical computer science; Combinatorics; Artificial intelligence; Physics","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009793334,0.0006115247,0.000896325,0.0002754377,0.0003897482,0.001082971,0.001168233,0.001173183,0.003250234],"category_scores_gemma":[0.004917146,0.000342989,0.0003005321,0.0004482763,0.0011926,0.001378656,0.001293312,0.00127477,0.0006065764],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001489317,"about_ca_system_score_gemma":0.001634062,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004752589,"about_ca_topic_score_gemma":0.004737149,"domain_scores_codex":[0.9993944,0.0001999518,0.00002300183,0.0001396931,0.0001451455,0.00009784653],"domain_scores_gemma":[0.998827,0.0005792013,0.00009346439,0.0001854091,0.0002368516,0.00007807038],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002075095,0.00005384216,0.0007792793,0.00009668907,0.00003523994,0.0001257535,0.00008182357,0.8748587,0.002133836,0.0844631,0.005287751,0.03187649],"study_design_scores_gemma":[0.000013713,0.0000162336,0.00005501264,0.0000037982,0.000002299622,0.000009007052,0.000006446769,0.9859035,0.0002924167,0.01314322,0.000551373,0.000002993055],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02372772,0.0004015529,0.96828,0.000792417,0.0001375925,0.00005070412,0.0001524367,0.000552397,0.005905148],"genre_scores_gemma":[0.8089463,0.0003152645,0.1804537,0.000488643,0.00009005103,0.0001369421,0.0003379812,0.000145056,0.009086006],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004752589,"threshold_uncertainty_score":0.01087314,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01959072728710984,"score_gpt":0.2404384011253772,"score_spread":0.2208476738382674,"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."}}