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

TL;DR

This paper introduces a novel ANN-SNN conversion method that explicitly models residual errors as noise, significantly improving SNN accuracy and efficiency under ultra-low-latency conditions for neuromorphic hardware applications.

Contribution

The paper presents a new residual error modeling approach that enhances ANN-SNN conversion, outperforming existing methods in accuracy and latency on standard datasets.

Findings

Outperforms existing conversion methods in accuracy.

Reduces required time steps for SNN inference.

Enhances practical deployment of SNNs on edge devices.

Abstract

Spiking neural networks (SNNs) have garnered interest due to their energy efficiency and superior effectiveness on neuromorphic chips compared with traditional artificial neural networks (ANNs). One of the mainstream approaches to implementing deep SNNs is the ANN-SNN conversion, which integrates the efficient training strategy of ANNs with the energy-saving potential and fast inference capability of SNNs. However, under extreme low-latency conditions, the existing conversion theory suggests that the problem of misrepresentation of residual membrane potentials in SNNs, i.e., the inability of IF neurons with a reset-by-subtraction mechanism to respond to residual membrane potentials beyond the range from resting potential to threshold, leads to a performance gap in the converted SNNs compared to the original ANNs. This severely limits the possibility of practical application of SNNs on delay-sensitive edge devices. Existing conversion methods addressing this problem usually involve modifying the state of the conversion spiking neurons. However, these methods do not consider their adaptability and compatibility with neuromorphic chips. We propose a new approach based on explicit modeling of residual errors as additive noise. The noise is incorporated into the activation function of the source ANN, which effectively reduces the residual error. Our experiments on the CIFAR10/100 dataset verify that our approach exceeds the prevailing ANN-SNN conversion methods and directly trained SNNs concerning accuracy and the required time steps. Overall, our method provides new ideas for improving SNN performance under ultra-low-latency conditions and is expected to promote practical neuromorphic hardware applications for further development.