Direct Training for Spiking Neural Networks: Faster, Larger, Better

TL;DR

This paper introduces a novel direct training method for deep spiking neural networks, achieving faster training, higher accuracy on neuromorphic datasets, and competitive performance on standard datasets, thus advancing the practical deployment of SNNs.

Contribution

It proposes a neuron normalization technique and an explicit iterative LIF model conversion, enabling efficient training of large-scale deep SNNs with significant speedup and improved accuracy.

Findings

Achieved higher accuracy on neuromorphic datasets (N-MNIST, DVS-CIFAR10).

Enabled training of deep SNNs with tens of times speedup.

Demonstrated competitive performance on CIFAR10 compared to ANNs.

Abstract

Spiking neural networks (SNNs) that enables energy efficient implementation on emerging neuromorphic hardware are gaining more attention. Yet now, SNNs have not shown competitive performance compared with artificial neural networks (ANNs), due to the lack of effective learning algorithms and efficient programming frameworks. We address this issue from two aspects: (1) We propose a neuron normalization technique to adjust the neural selectivity and develop a direct learning algorithm for deep SNNs. (2) Via narrowing the rate coding window and converting the leaky integrate-and-fire (LIF) model into an explicitly iterative version, we present a Pytorch-based implementation method towards the training of large-scale SNNs. In this way, we are able to train deep SNNs with tens of times speedup. As a result, we achieve significantly better accuracy than the reported works on neuromorphic datasets (N-MNIST and DVS-CIFAR10), and comparable accuracy as existing ANNs and pre-trained SNNs on non-spiking datasets (CIFAR10). {To our best knowledge, this is the first work that demonstrates direct training of deep SNNs with high performance on CIFAR10, and the efficient implementation provides a new way to explore the potential of SNNs.