{"id":"W2618636670","doi":"10.1016/j.parco.2017.05.003","title":"Proteus: Exploiting precision variability in deep neural networks","year":2017,"lang":"en","type":"article","venue":"Parallel Computing","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of British Columbia; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Memory footprint; Inference; Convolutional neural network; Artificial neural network; Proteus; Efficient energy use; Floating point; Computer engineering; Real-time computing; Algorithm; Artificial intelligence","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00158283,0.001102275,0.0008714188,0.0007841883,0.0005360073,0.001211244,0.002067271,0.000941433,0.003595531],"category_scores_gemma":[0.005365978,0.000772165,0.0006116833,0.0009903867,0.0006316295,0.002387444,0.001981676,0.001906466,0.0009912362],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007595967,"about_ca_system_score_gemma":0.001397347,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004853653,"about_ca_topic_score_gemma":0.008671984,"domain_scores_codex":[0.9993786,0.0001611011,0.00002713546,0.0001415906,0.0002203762,0.00007118137],"domain_scores_gemma":[0.9988092,0.000539218,0.00007667605,0.0003295037,0.0001801558,0.00006526878],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001111738,0.0002984697,0.003457347,0.0002939188,0.0003334505,0.0001724676,0.000155594,0.4314687,0.01622213,0.03272056,0.0258889,0.4878767],"study_design_scores_gemma":[0.00003381399,0.00004457917,0.0001722292,0.000008091904,0.00001455009,0.00001443084,0.000007925046,0.9819676,0.003408981,0.01308923,0.001230139,0.000008361995],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09607545,0.001347072,0.8671866,0.0006035143,0.0004647018,0.00008603714,0.000967367,0.02795463,0.005314589],"genre_scores_gemma":[0.6078084,0.0003874828,0.3819358,0.0003486695,0.0001713117,0.0001721998,0.001586755,0.002637075,0.004952356],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004853653,"threshold_uncertainty_score":0.01202828,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02995690329559604,"score_gpt":0.3004231432261359,"score_spread":0.2704662399305399,"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."}}