Spikemax: Spike-based Loss Methods for Classification

TL;DR

This paper introduces spike-based loss functions for training spiking neural networks, improving classification performance, energy efficiency, and inference speed on neuromorphic datasets.

Contribution

It formulates a novel spike-based negative log-likelihood loss for SNNs, enhancing compatibility with spike outputs and achieving state-of-the-art results.

Findings

Achieved faster inference speeds on benchmark datasets.

Reduced energy consumption in SNN classification.

Outperformed existing methods in accuracy on neuromorphic datasets.

Abstract

Spiking Neural Networks~(SNNs) are a promising research paradigm for low power edge-based computing. Recent works in SNN backpropagation has enabled training of SNNs for practical tasks. However, since spikes are binary events in time, standard loss formulations are not directly compatible with spike output. As a result, current works are limited to using mean-squared loss of spike count. In this paper, we formulate the output probability interpretation from the spike count measure and introduce spike-based negative log-likelihood measure which are more suited for classification tasks especially in terms of the energy efficiency and inference latency. We compare our loss measures with other existing alternatives and evaluate using classification performances on three neuromorphic benchmark datasets: NMNIST, DVS Gesture and N-TIDIGITS18. In addition, we demonstrate state of the art performances on these datasets, achieving faster inference speed and less energy consumption.