Abstract:Spiking Neural Networks (SNNs) operate with asynchronous discrete events (or spikes) which can potentially lead to higher energy-efficiency in neuromorphic hardware implementations. Many works have shown that an SNN for inference can be formed by copying the weights from a trained Artificial Neural Network (ANN) and setting the firing threshold for each layer as the maximum input received in that layer. These type of converted SNNs require a large number of time steps to achieve competitive accuracy which diminishes the energy savings. The number of time steps can be reduced by training SNNs with spike-based backpropagation from scratch, but that is computationally expensive and slow. To address these challenges, we present a computationally-efficient training technique for deep SNNs. We propose a hybrid training methodology: 1) take a converted SNN and use its weights and thresholds as an initialization step for spike-based backpropagation, and 2) perform incremental spike-timing dependent backpropagation (STDB) on this carefully initialized network to obtain an SNN that converges within few epochs and requires fewer time steps for input processing. STDB is performed with a novel surrogate gradient function defined using neuron's spike time. The proposed training methodology converges in less than 20 epochs of spike-based backpropagation for most standard image classification datasets, thereby greatly reducing the training complexity compared to training SNNs from scratch. We perform experiments on CIFAR-10, CIFAR-100, and ImageNet datasets for both VGG and ResNet architectures. We achieve top-1 accuracy of 65.19% for ImageNet dataset on SNN with 250 time steps, which is 10X faster compared to converted SNNs with similar accuracy.

A Little Energy Goes a Long Way: Build an Energy-Efficient, Accurate Spiking Neural Network From Convolutional Neural Network

BitSNNs: Revisiting Energy-efficient Spiking Neural Networks

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

Low Latency Spiking ConvNets with Restricted Output Training and False Spike Inhibition

Spiking Deep Residual Networks.

Toward High-Accuracy and Low-Latency Spiking Neural Networks With Two-Stage Optimization

Spiking Deep Convolutional Neural Networks for Energy-Efficient Object Recognition

Efficient and Accurate Conversion of Spiking Neural Network with Burst Spikes

Training a General Spiking Neural Network with Improved Efficiency and Minimum Latency

CSNN: an Augmented Spiking Based Framework with Perceptron-Inception

SpikeConverter: an Efficient Conversion Framework Zipping the Gap Between Artificial Neural Networks and Spiking Neural Networks

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

C-DNN: A 24.5-85.8TOPS/W Complementary-Deep-Neural-Network Processor with Heterogeneous CNN/SNN Core Architecture and Forward-Gradient-Based Sparsity Generation

SpikeAtConv: An Integrated Spiking-Convolutional Attention Architecture for Energy-Efficient Neuromorphic Vision Processing

Reconsidering the energy efficiency of spiking neural networks

STCSNN: High energy efficiency spike-train level spiking neural networks with spatio-temporal conversion

Towards High-performance Spiking Transformers from ANN to SNN Conversion

Optimizing Power Efficiency: Converting Recurrent Neural Networks to Spiking Neural Networks for Time-Domain Analysis

A Time-to-first-spike Coding and Conversion Aware Training for Energy-Efficient Deep Spiking Neural Network Processor Design

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

Homeostasis-Based Cnn-To-Snn Conversion Of Inception And Residual Architectures