Statistical Inference for Autoencoder-based Anomaly Detection after Representation Learning-based Domain Adaptation

TL;DR

This paper introduces STAND-DA, a framework that applies statistical inference to autoencoder-based anomaly detection after domain adaptation, ensuring valid p-values and controlled false positives in large-scale deep learning contexts.

Contribution

It develops a GPU-accelerated selective inference method for autoencoder-based anomaly detection post domain adaptation, enabling statistically valid conclusions in complex deep models.

Findings

Valid p-values are computed for anomalies.

False positive rate is rigorously controlled.

Method is scalable and efficient on large datasets.

Abstract

Anomaly detection (AD) plays a vital role across a wide range of domains, but its performance might deteriorate when applied to target domains with limited data. Domain Adaptation (DA) offers a solution by transferring knowledge from a related source domain with abundant data. However, this adaptation process can introduce additional uncertainty, making it difficult to draw statistically valid conclusions from AD results. In this paper, we propose STAND-DA -- a novel framework for statistically rigorous Autoencoder-based AD after Representation Learning-based DA. Built on the Selective Inference (SI) framework, STAND-DA computes valid $p$-values for detected anomalies and rigorously controls the false positive rate below a pre-specified level $\alpha$ (e.g., 0.05). To address the computational challenges of applying SI to deep learning models, we develop the GPU-accelerated SI implementation, significantly enhancing both scalability and runtime performance. This advancement makes SI practically feasible for modern, large-scale deep architectures. Extensive experiments on synthetic and real-world datasets validate the theoretical results and computational efficiency of the proposed STAND-DA method.