Energy-Efficient Convolutional Neural Networks with Deterministic Bit-Stream Processing
S. Rasoul Faraji1,a, M. Hassan Najafi2, Bingzhe Li1,b, David J. Lilja1,c and Kia Bazargan1,d
1University of Minnesota, Minneapolis, USA
afaraj008@umn.edu
blixx1743@umn.edu
clilja@umn.edu
dkia@umn.edu
2University of Louisiana at Lafayette, Lafayette, LA, USA
najafi@louisiana.edu
ABSTRACT
Stochastic computing (SC) has been used for lowcost and low power implementation of neural networks. Inherent inaccuracy and long latency of processing random bit-streams have made prior SC-based implementations inefficient compared to conventional fixed-point designs. Random or pseudo-random bitstreams often need to be processed for a very long time to produce acceptable results. This long latency leads to a significantly higher energy consumption than binary design counterparts. Low-discrepancy sequences have been recently used for fast-converging deterministic computation with stochastic constructs. In this work, we propose a low-cost, low-latency, and energy-efficient implementation of convolutional neural networks based on low-discrepancy deterministic bit-streams. Experimental results show a significant reduction in the energy consumption compared to previous random bitstream-based implementations and to the optimized fixed-point design with no quality degradation.
Keywords: Convolutional neural networks, Bitstream processing, Stochastic computing, Energy-efficient design, Low-cost design.
