Algorithm and Hardware Design of Discrete-Time Spiking Neural Networks Based on Back Propagation with Binary Activations
This work addresses the training challenge for SNNs, which are important for energy-efficient neuromorphic computing, though it appears incremental by adapting binarized neural network techniques.
The authors tackled the difficulty of training discrete-time spiking neural networks (SNNs) using back propagation by introducing a new algorithm based on binary activations, achieving high classification accuracy (>98% on MNIST) and low energy consumption (48.4-773 nJ/image) in neuromorphic hardware.
We present a new back propagation based training algorithm for discrete-time spiking neural networks (SNN). Inspired by recent deep learning algorithms on binarized neural networks, binary activation with a straight-through gradient estimator is used to model the leaky integrate-fire spiking neuron, overcoming the difficulty in training SNNs using back propagation. Two SNN training algorithms are proposed: (1) SNN with discontinuous integration, which is suitable for rate-coded input spikes, and (2) SNN with continuous integration, which is more general and can handle input spikes with temporal information. Neuromorphic hardware designed in 40nm CMOS exploits the spike sparsity and demonstrates high classification accuracy (>98% on MNIST) and low energy (48.4-773 nJ/image).