{"id":"W2786618695","doi":"10.1109/fpt.2017.8280150","title":"FPGA-based training of convolutional neural networks with a reduced precision floating-point library","year":2017,"lang":"en","type":"article","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"MNIST database; Adder; Convolutional neural network; Computer science; Field-programmable gate array; Floating point; Double-precision floating-point format; Lookup table; IEEE floating point; Single-precision floating-point format; Artificial neural network; Artificial intelligence; Bandwidth (computing); Computer hardware; Parallel computing; Algorithm; Latency (audio)","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002309185,0.0005698169,0.0001876458,0.0003944914,0.0001920705,0.0004219066,0.001068399,0.0002429255,0.006616543],"category_scores_gemma":[0.000761693,0.0002356383,0.0002105353,0.0003419604,0.000201545,0.0005517905,0.0002243276,0.0004428139,0.001346335],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006297916,"about_ca_system_score_gemma":0.0008436205,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004319425,"about_ca_topic_score_gemma":0.008260953,"domain_scores_codex":[0.9998558,0.00001680154,0.00001091619,0.00003280472,0.00005360852,0.00003003285],"domain_scores_gemma":[0.9997975,0.00007083799,0.0000226825,0.00004445934,0.00005566153,0.000008871794],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0008677511,0.0003130289,0.005456622,0.0007212473,0.0001835835,0.0006728214,0.0002094713,0.2711836,0.1458761,0.01545306,0.02170745,0.5373552],"study_design_scores_gemma":[0.00008933628,0.0003529597,0.001915439,0.00005246132,0.0000434182,0.0002524576,0.00003711291,0.7844212,0.1950452,0.001739966,0.01602643,0.00002390607],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.3811594,0.0009442354,0.5568738,0.0003677982,0.0002410862,0.0002451119,0.0009411046,0.02759133,0.03163612],"genre_scores_gemma":[0.7701506,0.0002988002,0.2173949,0.0001814728,0.0000196675,0.0001436265,0.001131118,0.0003820082,0.01029775],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006616543,"threshold_uncertainty_score":0.02213448,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03390554162329246,"score_gpt":0.2668299662184518,"score_spread":0.2329244245951593,"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."}}