Parallel implementation of neural networks training on graphic processing unit

Author

Yong Liu ; Yeming Xiao ; Li Wang ; Jielin Pan ; Yonghong Yan

Author_Institution

Key Lab. of Speech Acoust. & Content Understanding, Inst. of Acoust., Beijing, China

fYear

2012

fDate

16-18 Oct. 2012

Firstpage

1571

Lastpage

1574

Abstract

Recently artificial neural network (ANN) especially the deep belief network (DBN) becomes more and more popular in the acoustic model training. In order to improve the speed of ANN, the Graphics Processing Unit (GPU) is used. This paper gives the training details of the Back-Propagation (BP) neural network acoustic model for speech recognition on GPU, including the parallel reduction application and asynchronous implementation between CPU and GPU. It is 26 times faster than using the single thread Intel^® MKL(Math Kernel Library) implementation.

Keywords

acoustic signal processing; backpropagation; belief networks; graphics processing units; neural nets; parallel programming; speech recognition; ANN; CPU; DBN; GPU; acoustic model training; artificial neural network; asynchronous implementation; backpropagation neural network acoustic model; deep belief network; graphic processing unit; neural network training; parallel implementation; parallel reduction application; speech recognition; BP neural network; GPU; acoustic model; speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Biomedical Engineering and Informatics (BMEI), 2012 5th International Conference on

Conference_Location

Chongqing

Print_ISBN

978-1-4673-1183-0

Type

conf

DOI

10.1109/BMEI.2012.6513078

Filename

6513078