Unsupervised Domain Shift Detection with Interpretable Subspace Attribution

TL;DR

This paper introduces an interpretable tool for detecting and understanding subtle domain shifts in high-dimensional data, enabling identification of specific features responsible for distributional differences.

Contribution

The authors present a novel algorithm that detects local density anomalies and identifies feature subspaces, improving interpretability of domain shifts in high-dimensional datasets.

Findings

Successfully detects both broad and localized shifts in benchmarks.

Identifies device-induced shifts in ECG data and associated features.

Provides a protocol for removing residual distributional differences.

Abstract

We developed a tool for detecting domain shifts, namely subtle differences in the probability distributions of datasets. We identify these shifts using an algorithm designed to detect localised density anomalies in high-dimensional feature spaces. If an anomaly is present, we then identify the feature subspace in which the anomaly is most pronounced. This allows us to trace the domain shift to a small set of features, making the shift interpretable. Moreover, we provide a protocol for compensating domain shifts by extracting, from two unlabelled datasets, subsets of samples with no detectable residual distributional difference. We validate the framework on controlled 20-dimensional benchmarks with known ground truth, recovering both broad and localized shifts together with their supporting feature subspaces. We then apply it to healthy electrocardiogram (ECG) recordings represented by 782 features. In age- and sex-matched cohort comparisons differing in measurement-device composition, the method detects device-induced shifts, extracts representative subsets enriched in the imbalanced device components, and identifies ECG features associated with the acquisition contrast. These results suggest that density-shift detection and subspace attribution provide a practical framework for uncovering hidden cohort biases before downstream modelling.