Unleashing the Potential of Unsupervised Deep Outlier Detection through Automated Training Stopping

TL;DR

This paper introduces a novel, label-free method for automatically determining the optimal training stopping point in deep outlier detection models, enhancing robustness, performance, and training efficiency across diverse datasets.

Contribution

It proposes a new loss entropy metric and an automated stopping algorithm that reliably identifies the best training iteration without labels, addressing hyperparameter sensitivity issues.

Findings

Improves robustness of deep OD models to hyperparameter variations

Reduces training time significantly compared to naive training

Enhances model performance across diverse datasets

Abstract

Outlier detection (OD) has received continuous research interests due to its wide applications. With the development of deep learning, increasingly deep OD algorithms are proposed. Despite the availability of numerous deep OD models, existing research has reported that the performance of deep models is extremely sensitive to the configuration of hyperparameters (HPs). However, the selection of HPs for deep OD models remains a notoriously difficult task due to the lack of any labels and long list of HPs. In our study. we shed light on an essential factor, training time, that can introduce significant variation in the performance of deep model. Even the performance is stable across other HPs, training time itself can cause a serious HP sensitivity issue. Motivated by this finding, we are dedicated to formulating a strategy to terminate model training at the optimal iteration. Specifically, we propose a novel metric called loss entropy to internally evaluate the model performance during training while an automated training stopping algorithm is devised. To our knowledge, our approach is the first to enable reliable identification of the optimal training iteration during training without requiring any labels. Our experiments on tabular, image datasets show that our approach can be applied to diverse deep models and datasets. It not only enhances the robustness of deep models to their HPs, but also improves the performance and reduces plenty of training time compared to naive training.