Abstract:Spiking neural networks (SNNs) can do inference with low power consumption due to their spike sparsity. Although SNNs can be combined with neuromorphic hardware to achieve efficient inference, they are often difficult to train directly due to discrete non-differentiable spikes. As an alternative, ANN-SNN conversion is an efficient way to achieve deep SNNs by converting well-trained artificial neural networks (ANNs). However, the existing methods commonly use constant threshold for conversion. A high constant threshold value prevents neurons from rapidly delivering spikes to deeper layers and causes high time delay. In addition, the same response for different inputs may result in information loss during the information transmission. Inspired by the biological adaptive threshold mechanism, we propose a multistage adaptive threshold (MSAT) method to alleviate this problem. Instead of using a single, constant value, the threshold is adjusted in multistages, adapting to each neuron’s firing history and input properties. Specifically, for each neuron, the dynamic threshold is positively correlated with the average membrane potential and negatively correlated with the rate of depolarization. The adaptation to membrane potential and input allows a timely adjustment of the threshold to fire spikes faster and transmit more information. Moreover, we analyze the spikes of inactivated neurons error, which is pervasive in early time steps. We also propose spike confidence accordingly to measure confidence about the neurons that correctly deliver spikes. Such spike confidence in early time steps is used to determine whether to elicit the spike to alleviate the spikes of inactivated neurons error. Combined with the proposed methods, we examine the performance on CIFAR-10, CIFAR-100, and ImageNet datasets. We also conduct sentiment classification and speech recognition experiments on the IDBM and Google speech commands datasets, respectively. Experiments show that our methods can achieve near-lossless and lower latency ANN-SNN conversion. In summary, we build a biologically inspired multistage adaptive threshold for converted SNN, with comparable performance to state-of-the-art methods while improving energy efficiency.

CIRM-SNN: Certainty Interval Reset Mechanism Spiking Neuron for Enabling High Accuracy Spiking Neural Network

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

Efficient and Accurate Conversion of Spiking Neural Network with Burst Spikes

Optimized Potential Initialization for Low-latency Spiking Neural Networks

Event-driven Spiking Neural Network Based on Membrane Potential Modulation for Remote Sensing Image Classification

A New ANN-SNN Conversion Method with High Accuracy, Low Latency and Good Robustness

MSAT: biologically inspired multistage adaptive threshold for conversion of spiking neural networks

Spike Calibration: Bridging the Gap between ANNs and SNNs in ANN-SNN Conversion

Reducing ANN-SNN Conversion Error Through Residual Membrane Potential.

Converting High-Performance and Low-Latency SNNs through Explicit Modelling of Residual Error in ANNs

Spiking Deep Residual Networks.

High-Accuracy Spiking Neural Network for Objective Recognition Based on Proportional Attenuating Neuron

An adaptive threshold mechanism for accurate and efficient deep spiking convolutional neural networks

Optimal ANN-SNN Conversion for High-accuracy and Ultra-low-latency Spiking Neural Networks

Bridging the Gap between ANNs and SNNs by Calibrating Offset Spikes

Optimal ANN-SNN Conversion for Fast and Accurate Inference in Deep Spiking Neural Networks

MSAT: Biologically Inspired Multi-Stage Adaptive Threshold for Conversion of Spiking Neural Networks

Optimal Conversion of Conventional Artificial Neural Networks to Spiking Neural Networks

A universal ANN-to-SNN framework for achieving high accuracy and low latency deep Spiking Neural Networks

Spiking Neural Networks with Consistent Mapping Relations Allow High-Accuracy Inference

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