Neural architecture search for energy-efficient always-on audio machine learning

• Published in: Neural computing & applications 2023-06, Vol.35 (16), p.12133-12144
• Main Authors: Speckhard, Daniel T., Misiunas, Karolis, Perel, Sagi, Zhu, Tenghui, Carlile, Simon, Slaney, Malcolm
• Format: Article
• Language: English
• Subjects: Accuracy; Artificial Intelligence; Classification; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer architecture; Computer Science; Data Mining and Knowledge Discovery; Edge computing; Energy consumption; Energy efficiency; Hardware; Image Processing and Computer Vision; Machine learning; Mobile computing; Neural networks; Original Article; Probability and Statistics in Computer Science; Search methods; Sound filters
• ISSN: 0941-0643; 1433-3058
• DOI: 10.1007/s00521-023-08345-y

Mobile and edge computing devices for always-on classification tasks require energy-efficient neural network architectures. In this paper we present several changes to neural architecture searches that improve the chance of success in practical situations. Our search simultaneously optimizes for network accuracy, energy efficiency and memory usage. We benchmark the performance of our search on real hardware, but since running thousands of tests with real hardware is difficult, we use a random forest model to roughly predict the energy usage of a candidate network. We present a search strategy that uses both Bayesian and regularized evolutionary search with particle swarms, and employs early stopping to reduce the computational burden. Our search, evaluated on a sound event classification dataset based upon AudioSet, results in an order of magnitude less energy per inference and a much smaller memory footprint than our baseline MobileNetV1/V2 implementations while slightly improving task accuracy. We also demonstrate how combining a 2D spectrogram with a convolution with many filters causes a computational bottleneck for audio classification and that alternative approaches reduce the computational burden but sacrifice task accuracy.