SLAYER: Spike Layer Error Reassignment in Time

TL;DR

This paper introduces SLAYER, a novel backpropagation method for training deep Spiking Neural Networks that overcomes non-differentiability issues, enabling state-of-the-art performance on multiple datasets.

Contribution

SLAYER presents a new general backpropagation algorithm for SNNs that handles non-differentiable spike functions and incorporates temporal credit assignment.

Findings

Achieves state-of-the-art results on MNIST, NMNIST, DVS Gesture, TIDIGITS datasets.

Provides GPU-accelerated software for training SNNs.

Outperforms existing SNN learning methods and conversion techniques.

Abstract

Configuring deep Spiking Neural Networks (SNNs) is an exciting research avenue for low power spike event based computation. However, the spike generation function is non-differentiable and therefore not directly compatible with the standard error backpropagation algorithm. In this paper, we introduce a new general backpropagation mechanism for learning synaptic weights and axonal delays which overcomes the problem of non-differentiability of the spike function and uses a temporal credit assignment policy for backpropagating error to preceding layers. We describe and release a GPU accelerated software implementation of our method which allows training both fully connected and convolutional neural network (CNN) architectures. Using our software, we compare our method against existing SNN based learning approaches and standard ANN to SNN conversion techniques and show that our method achieves state of the art performance for an SNN on the MNIST, NMNIST, DVS Gesture, and TIDIGITS datasets.