OutlierNets: Highly Compact Deep Autoencoder Network Architectures for On-Device Acoustic Anomaly Detection

TL;DR

OutlierNets introduces highly compact deep autoencoder architectures for acoustic anomaly detection, enabling efficient on-device deployment with minimal computational resources while maintaining high detection accuracy.

Contribution

The paper presents a novel design exploration strategy that produces extremely compact autoencoder networks, significantly reducing size and latency without sacrificing accuracy.

Findings

Models as small as 2.7 KB achieve comparable accuracy to larger architectures.

OutlierNets can be up to 21 times faster in latency on CPU.

Networks require as few as 686 parameters and 2.8 million FLOPs.

Abstract

Human operators often diagnose industrial machinery via anomalous sounds. Automated acoustic anomaly detection can lead to reliable maintenance of machinery. However, deep learning-driven anomaly detection methods often require an extensive amount of computational resources which prohibits their deployment in factories. Here we explore a machine-driven design exploration strategy to create OutlierNets, a family of highly compact deep convolutional autoencoder network architectures featuring as few as 686 parameters, model sizes as small as 2.7 KB, and as low as 2.8 million FLOPs, with a detection accuracy matching or exceeding published architectures with as many as 4 million parameters. Furthermore, CPU-accelerated latency experiments show that the OutlierNet architectures can achieve as much as 21x lower latency than published networks.