An Experimental Analysis of the Power Consumption of Convolutional Neural Networks for Keyword Spotting

TL;DR

This paper empirically investigates the power consumption of convolutional neural networks used for keyword spotting on a Raspberry Pi, highlighting the predictive value of multiply operations for energy use.

Contribution

It provides an experimental analysis linking model complexity measures to actual power consumption in real-world deployment scenarios.

Findings

Multiply operations are better predictors of energy consumption than model parameters.

Higher accuracy models tend to consume more power.

Proxy measures like parameter count are useful but less predictive of energy use.

Abstract

Nearly all previous work on small-footprint keyword spotting with neural networks quantify model footprint in terms of the number of parameters and multiply operations for a feedforward inference pass. These values are, however, proxy measures since empirical performance in actual deployments is determined by many factors. In this paper, we study the power consumption of a family of convolutional neural networks for keyword spotting on a Raspberry Pi. We find that both proxies are good predictors of energy usage, although the number of multiplies is more predictive than the number of model parameters. We also confirm that models with the highest accuracies are, unsurprisingly, the most power hungry.