{"id":"W3120626278","doi":"10.1109/jiot.2021.3051611","title":"A Novel Adaptive Gradient Compression Scheme: Reducing the Communication Overhead for Distributed Deep Learning in the Internet of Things","year":2021,"lang":"en","type":"article","venue":"IEEE Internet of Things Journal","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia; Carleton University","funders":"Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Computer science; Overhead (engineering); Deep learning; Data compression; Artificial intelligence; Node (physics); Compression ratio; Edge computing; Artificial neural network; Transmission (telecommunications); Distributed computing; Machine learning; Computer network; Computer engineering; Enhanced Data Rates for GSM Evolution; Telecommunications","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":[0.001092843,0.0001498201,0.0002457135,0.00007924195,0.0001717325,0.0001422259,0.00210287,0.00006150416,0.000004412586],"category_scores_gemma":[0.0002867681,0.00009944617,0.0001563361,0.0004272279,0.000136192,0.0007474578,0.0004078959,0.001004929,8.170376e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001046291,"about_ca_system_score_gemma":0.00004911583,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00012938,"about_ca_topic_score_gemma":0.00001429337,"domain_scores_codex":[0.99822,0.0002576745,0.000666222,0.000245827,0.0003701872,0.00024013],"domain_scores_gemma":[0.997073,0.0009943477,0.0009519696,0.0005475765,0.0003868314,0.00004628106],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006824567,0.001848212,0.001541485,0.0001820299,0.0004989275,0.00003549285,0.2369886,0.08233009,0.3440641,0.1409479,0.008047949,0.1828327],"study_design_scores_gemma":[0.0007291533,0.0001726628,0.0006370255,0.0008676653,0.00002064363,0.0003947596,0.001398378,0.9453687,0.03958269,0.009006134,0.001672277,0.0001498909],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08291837,0.0005333923,0.913586,0.002413001,0.0001942333,0.000230832,0.000001837078,0.00002147028,0.0001008884],"genre_scores_gemma":[0.8886147,0.00007618828,0.1108469,0.0003184016,0.00003772372,0.00002554001,0.00001011273,0.00001058999,0.00005992462],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8630387,"threshold_uncertainty_score":0.4365971,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03464347156434745,"score_gpt":0.2928884088350212,"score_spread":0.2582449372706738,"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."}}