pdlADMM: An ADMM-based framework for parallel deep learning training with efficiency
Sign Up to like & getrecommendations! Published in 2021 at "Neurocomputing"
DOI: 10.1016/j.neucom.2020.09.029
Abstract: Abstract Alternating Direction Methods of Multipliers (ADMM) has been proven to be a useful alternative to the popular gradient-based optimizers and successfully applied to train the DNN model. Whereas existing ADMM-based approaches generally do not… read more here.
Keywords: based framework; parallel deep; framework parallel; pdladmm ... See more keywords