Unsupervised deep learning techniques are widely used to identify anomalous behaviour. The performance of such methods is a product of the amount of training data and the model size. However, the size is often a limiting factor for the deployment on resource-constrained devices. We present a novel procedure based on knowledge distillation for compressing an unsupervised anomaly detection model into a supervised deployable one and we suggest a set of techniques to improve the detection sensitivity. Compressed models perform comparably to their larger counterparts while significantly reducing the size and memory footprint.
@article{arxiv.2310.06047,
title = {Knowledge Distillation for Anomaly Detection},
author = {Adrian Alan Pol and Ekaterina Govorkova and Sonja Gronroos and Nadezda Chernyavskaya and Philip Harris and Maurizio Pierini and Isobel Ojalvo and Peter Elmer},
journal= {arXiv preprint arXiv:2310.06047},
year = {2023}
}