RMP-SNN: Residual Membrane Potential Neuron for Enabling Deeper High-Accuracy and Low-Latency Spiking Neural Network

TLDR

Spiking neural networks, the third generation of artificial neural networks, promise low‑power event‑driven analytics, yet state‑of‑the‑art image‑recognition SNNs are typically obtained by converting trained ReLU‑based ANNs, a process that incurs accuracy loss and demands many inference time‑steps. This work proposes converting ANNs to SNNs with a soft‑reset Residual Membrane Potential neuron that preserves residual membrane potential at firing. The conversion replaces hard‑reset integrate‑and‑fire neurons with a soft‑reset RMP neuron that retains the membrane potential above threshold during firing, thereby reducing information loss. Experiments on VGG‑16, ResNet‑20, and ResNet‑34 show near loss‑less conversion, achieving 93.63 % top‑1 on CIFAR‑10, 70.93 % on CIFAR‑100, and 73.09 % on ImageNet, while outperforming hard‑reset SNNs with 2–8× fewer inference steps.

Abstract

Spiking Neural Networks (SNNs) have recently attracted significant research interest as the third generation of artificial neural networks that can enable low-power event-driven data analytics. The best performing SNNs for image recognition tasks are obtained by converting a trained Analog Neural Network (ANN), consisting of Rectified Linear Units (ReLU), to SNN composed of integrate-and-fire neurons with "proper" firing thresholds. The converted SNNs typically incur loss in accuracy compared to that provided by the original ANN and require sizable number of inference time-steps to achieve the best accuracy. We find that performance degradation in the converted SNN stems from using "hard reset" spiking neuron that is driven to fixed reset potential once its membrane potential exceeds the firing threshold, leading to information loss during SNN inference. We propose ANN-SNN conversion using "soft reset" spiking neuron model, referred to as Residual Membrane Potential (RMP) spiking neuron, which retains the "residual" membrane potential above threshold at the firing instants. We demonstrate near loss-less ANN-SNN conversion using RMP neurons for VGG-16, ResNet-20, and ResNet-34 SNNs on challenging datasets including CIFAR-10 (93.63% top-1), CIFAR-100 (70.93% top-1), and ImageNet (73.09% top-1 accuracy). Our results also show that RMP-SNN surpasses the best inference accuracy provided by the converted SNN with "hard reset" spiking neurons using 2-8 times fewer inference time-steps across network architectures and datasets.