Abstract
Human operators often diagnose industrial machinery via anomalous sounds. Given the new advances in the field of machine learning, 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 prohibiting 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. The architectures are deployed on an Intel Core i5 as well as a ARM Cortex A72 to assess performance on hardware that is likely to be used in industry. Experimental results on the model’s latency show that the OutlierNet architectures can achieve as much as 30× lower latency than published networks.
Highlights
Human operators often diagnose industrial machinery via anomalous sounds
A variety of deep learning-driven techniques have been introduced for acoustic anomaly detection (AAD) in recent years, including dense autoencoders [1,2], convolutional autoencoders [2], and pre-trained convolutional neural networks [3]
One of the primary reasons for the slow adoption is the prohibitive computational resources required by many deep learning-driven anomaly detection methods, which often have high architectural and computational complexities, as well as high memory footprints
Summary
Human operators often diagnose industrial machinery via anomalous sounds. Given the new advances in the field of machine learning, automated acoustic anomaly detection can lead to reliable maintenance of machinery. Acoustic anomalies are one of the primary ways through which malfunctioning machinery or industrial processes are monitored This detection of abnormal sounds is typically done subjectively via human operators who need prior experience. One of the primary reasons for the slow adoption is the prohibitive computational resources required by many deep learning-driven anomaly detection methods, which often have high architectural and computational complexities, as well as high memory footprints. This results in methods that cannot be deployed on resource-constrained edge computing devices such as CPUs or microcontrollers. Since such methods are typically designed without resource constraints, the on-device prediction latency is typically not an important design parameter
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
