Although Graphics Processing Units (GPUs) seem to currently be the best platform to train machine learning models, most research laboratories are still only equipped with standard CPU systems. In this paper, we investigate multiple techniques to speedup the training of Restricted Boltzmann Machine (RBM) models and Convolutional RBM (CRBM) models on CPU with the Contrastive Divergence (CD) algorithm. Experimentally, we show that the proposed techniques can reduce the training time by up to 30 times for RBM and up to 12 times for CRBM, on a data set of handwritten digits.
Détails
Titre
On CPU performance optimization of restricted Boltzmann machine and convolutional RBM
Auteur(s)/ trice(s)
Wicht, Baptiste (School of Engineering and Architecture (HEIA-FR), HES-SO University of Applied Sciences Western Switzerland ; University of Fribourg, Fribourg, Switzerland) Fischer, Andreas (School of Engineering and Architecture (HEIA-FR), HES-SO University of Applied Sciences Western Switzerland ; University of Fribourg, Fribourg, Switzerland) Hennebert, Jean (School of Engineering and Architecture (HEIA-FR), HES-SO University of Applied Sciences Western Switzerland ; University of Fribourg, Fribourg, Switzerland)
Date
2016-09
Publié dans
Proceedings of the 7th IAPR TC3 Workshop, Artificial Neural Networks in Pattern Recognition (ANNPR) 2016, 28-30 September 2016, Ulm Germany
Publié par
Ulm, Germany, 28-30 September 2016
Pagination
pp. 163-174
Présenté à
Artificial Neural Networks in Pattern Recognition, Ulm, Germany, 2016-09-28, 2016-09-30