Efficient Spiking Neural Networks with Radix Encoding

TL;DR

This paper introduces a radix encoding method for spiking neural networks that achieves ultra-short spike trains, significantly reducing latency and computational cost while maintaining high accuracy.

Contribution

The paper proposes a novel radix encoding scheme for SNNs that enables spike trains of less than ten steps, improving efficiency and speed over existing methods.

Findings

25X speedup compared to state-of-the-art

1.1% accuracy improvement on CIFAR-10

Spike trains under ten steps with maintained accuracy

Abstract

Spiking neural networks (SNNs) have advantages in latency and energy efficiency over traditional artificial neural networks (ANNs) due to its event-driven computation mechanism and replacement of energy-consuming weight multiplications with additions. However, in order to reach accuracy of its ANN counterpart, it usually requires long spike trains to ensure the accuracy. Traditionally, a spike train needs around one thousand time steps to approach similar accuracy as its ANN counterpart. This offsets the computation efficiency brought by SNNs because longer spike trains mean a larger number of operations and longer latency. In this paper, we propose a radix encoded SNN with ultra-short spike trains. In the new model, the spike train takes less than ten time steps. Experiments show that our method demonstrates 25X speedup and 1.1% increment on accuracy, compared with the state-of-the-art work on VGG-16 network architecture and CIFAR-10 dataset.