Optimizing the Consumption of Spiking Neural Networks with Activity Regularization

TL;DR

This paper explores methods to enhance energy efficiency in spiking neural networks by applying activity regularization techniques to enforce sparsity, thereby reducing synaptic operations and energy consumption on edge devices.

Contribution

It introduces and compares various activation regularization methods to optimize DNN-to-SNN conversion for energy efficiency without sacrificing accuracy.

Findings

Regularization improves sparsity in activation maps.

Enhanced sparsity reduces synaptic operations and energy use.

Different regularizers have varying impacts on efficiency.

Abstract

Reducing energy consumption is a critical point for neural network models running on edge devices. In this regard, reducing the number of multiply-accumulate (MAC) operations of Deep Neural Networks (DNNs) running on edge hardware accelerators will reduce the energy consumption during inference. Spiking Neural Networks (SNNs) are an example of bio-inspired techniques that can further save energy by using binary activations, and avoid consuming energy when not spiking. The networks can be configured for equivalent accuracy on a task through DNN-to-SNN conversion frameworks but their conversion is based on rate coding therefore the synaptic operations can be high. In this work, we look into different techniques to enforce sparsity on the neural network activation maps and compare the effect of different training regularizers on the efficiency of the optimized DNNs and SNNs.