A simple and efficient SNN and its performance & robustness evaluation method to enable hardware implementation

TL;DR

This paper introduces a simple, efficient 2-layer spiking neural network with improved hardware-friendly design, along with a novel performance and robustness evaluation method that does not rely on traditional recognition tests.

Contribution

It presents a lightweight SNN architecture optimized for hardware implementation and a new performance evaluation method that assesses accuracy and noise robustness efficiently.

Findings

Network uses fewer neurons and synapses with faster training.

Evaluation method achieves high correlation with recognition accuracy.

Robustness metric effectively measures noise tolerance.

Abstract

Spiking Neural Networks (SNN) are more closely related to brain-like computation and inspire hardware implementation. This is enabled by small networks that give high performance on standard classification problems. In literature, typical SNNs are deep and complex in terms of network structure, weight update rules and learning algorithms. This makes it difficult to translate them into hardware. In this paper, we first develop a simple 2-layered network in software which compares with the state of the art on four different standard data-sets within SNNs and has improved efficiency. For example, it uses lower number of neurons (3 x), synapses (3.5 x) and epochs for training (30 x) for the Fisher Iris classification problem. The efficient network is based on effective population coding and synapse-neuron co-design. Second, we develop a computationally efficient (15000 x) and accurate (correlation of 0.98) method to evaluate the performance of the network without standard recognition tests. Third, we show that the method produces a robustness metric that can be used to evaluate noise tolerance.