Mitigating Long-Tailed Anomaly Score Distributions with Importance-Weighted Loss

TL;DR

This paper introduces an importance-weighted loss function to address long-tailed anomaly score distributions in industrial anomaly detection, improving model robustness without prior class knowledge.

Contribution

It proposes a novel importance-weighted loss that mitigates long-tailed anomaly score distributions without needing prior normal class information.

Findings

Improves anomaly detection performance by 0.043 on benchmark datasets.

Effectively balances anomaly score distribution with a Gaussian target.

Demonstrates robustness across diverse real-world datasets.

Abstract

Anomaly detection is crucial in industrial applications for identifying rare and unseen patterns to ensure system reliability. Traditional models, trained on a single class of normal data, struggle with real-world distributions where normal data exhibit diverse patterns, leading to class imbalance and long-tailed anomaly score distributions (LTD). This imbalance skews model training and degrades detection performance, especially for minority instances. To address this issue, we propose a novel importance-weighted loss designed specifically for anomaly detection. Compared to the previous method for LTD in classification, our method does not require prior knowledge of normal data classes. Instead, we introduce a weighted loss function that incorporates importance sampling to align the distribution of anomaly scores with a target Gaussian, ensuring a balanced representation of normal data. Extensive experiments on three benchmark image datasets and three real-world hyperspectral imaging datasets demonstrate the robustness of our approach in mitigating LTD-induced bias. Our method improves anomaly detection performance by 0.043, highlighting its effectiveness in real-world applications.