Abstract:Spiking Neural Networks (SNNs) are a pathway that could potentially empower low-power event-driven neuromorphic hardware due to their spatio-temporal information processing capability and high biological plausibility. Although SNNs are currently more efficient than artificial neural networks (ANNs), they are not as accurate as ANNs. Error backpropagation is the most common method for directly training neural networks, promoting the prosperity of ANNs in various deep learning fields. However, since the signals transmitted in the SNN are non-differentiable discrete binary spike events, the activation function in the form of spikes presents difficulties for the gradient-based optimization algorithms to be directly applied in SNNs, leading to a performance gap (i.e., accuracy and latency) between SNNs and ANNs. This paper introduces a new learning algorithm, called SSTDP, which bridges the gap between backpropagation (BP)-based learning and spike-time-dependent plasticity (STDP)-based learning to train SNNs efficiently. The scheme incorporates the global optimization process from BP and the efficient weight update derived from STDP. It not only avoids the non-differentiable derivation in the BP process but also utilizes the local feature extraction property of STDP. Consequently, our method can lower the possibility of vanishing spikes in BP training and reduce the number of time steps to reduce network latency. In SSTDP, we employ temporal-based coding and use Integrate-and-Fire (IF) neuron as the neuron model to provide considerable computational benefits. Our experiments show the effectiveness of the proposed SSTDP learning algorithm on the SNN by achieving the best classification accuracy 99.3% on the Caltech 101 dataset, 98.1% on the MNIST dataset, and 91.3% on the CIFAR-10 dataset compared to other SNNs trained with other learning methods. It also surpasses the best inference accuracy of the directly trained SNN with 25~32× less inference latency. Moreover, we analyze event-based computations to demonstrate the efficacy of the SNN for inference operation in the spiking domain, and SSTDP methods can achieve 1.3~37.7× fewer addition operations per inference. The code is available at: https://github.com/MXHX7199/SNN-SSTDP .

Deep Spiking Neural Network with Ternary Spikes

An Energy Efficient Residual Spiking Neural Network Accelerator with Ternary Spikes

Ternary Spike: Learning Ternary Spikes for Spiking Neural Networks

Spike Trains Encoding and Threshold Rescaling Method for Deep Spiking Neural Networks

Spiking Deep Residual Networks.

Deep Spiking Convolutional Neural Networks for Programmable Neuro-synaptic System

Algorithm and hardware design of discrete-time spiking neural networks based on back propagation with binary activations

Deep Spiking Neural Network with Spike Count based Learning Rule

Deep Spiking Neural Networks with Binary Weights for Object Recognition

Hierarchical Spiking-Based Model for Efficient Image Classification with Enhanced Feature Extraction and Encoding.

Deep CovDenseSNN: A Hierarchical Event-Driven Dynamic Framework with Spiking Neurons in Noisy Environment

Training Energy-Efficient Deep Spiking Neural Networks with Single-Spike Hybrid Input Encoding

Enabling Deep Spiking Neural Networks with Hybrid Conversion and Spike Timing Dependent Backpropagation

An all integer-based spiking neural network with dynamic threshold adaptation

Temporal-Coded Deep Spiking Neural Network with Easy Training and Robust Performance.

High-performance deep spiking neural networks with 0.3 spikes per neuron

SSTDP: Supervised Spike Timing Dependent Plasticity for Efficient Spiking Neural Network Training

PT-Spike: A Precise-Time-Dependent Single Spike Neuromorphic Architecture with Efficient Supervised Learning

Nonpolar Neuron for ANN-SNN Conversion Toward Ternary Spiking Neural Network

Unsupervised Learning in a Ternary SNN Using STDP