From Computation to Consumption: Exploring the Compute-Energy Link for Training and Testing Neural Networks for SED Systems

TL;DR

This paper investigates the energy consumption of neural networks used in sound event detection, analyzing how factors like architecture size and GPU usage impact environmental costs.

Contribution

It provides a detailed analysis of energy requirements for training and testing neural networks in sound event detection, highlighting relationships with computational metrics.

Findings

Energy consumption correlates with model size and GPU utilization.

Complex relationships exist between energy use, floating-point operations, and parameters.

Insights can guide more energy-efficient neural network design.

Abstract

The massive use of machine learning models, particularly neural networks, has raised serious concerns about their environmental impact. Indeed, over the last few years we have seen an explosion in the computing costs associated with training and deploying these systems. It is, therefore, crucial to understand their energy requirements in order to better integrate them into the evaluation of models, which has so far focused mainly on performance. In this paper, we study several neural network architectures that are key components of sound event detection systems, using an audio tagging task as an example. We measure the energy consumption for training and testing small to large architectures and establish complex relationships between the energy consumption, the number of floating-point operations, the number of parameters, and the GPU/memory utilization.