Feature Shift Localization Network

2025-06-10

AI Summary

Feature distribution shift—prevalent in multi-source heterogeneous data (e.g., healthcare, finance, multi-sensor systems)—poses significant challenges for model generalization; existing approaches suffer from accuracy limitations and poor scalability to high-dimensional settings. Method: We propose the first end-to-end, generalizable neural framework that precisely localizes feature-level shifts across datasets and shift types—without requiring model retraining. Our approach integrates deep statistical modeling, meta-learning, and unsupervised feature importance estimation to explicitly characterize distributional discrepancies and quantify per-dimension shift magnitude. Contribution/Results: The framework achieves millisecond-scale inference on large-scale, high-dimensional data, significantly outperforming conventional statistical tests (e.g., KS, MMD) and state-of-the-art learning-based methods in localization accuracy. It is robust to diverse shift types—including covariate, concept, and label shift—and generalizes across domains. Source code and pre-trained models are publicly released.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Distributed Search
• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Feature shifts between data sources are present in many applications involving healthcare, biomedical, socioeconomic, financial, survey, and multi-sensor data, among others, where unharmonized heterogeneous data sources, noisy data measurements, or inconsistent processing and standardization pipelines can lead to erroneous features. Localizing shifted features is important to address the underlying cause of the shift and correct or filter the data to avoid degrading downstream analysis. While many techniques can detect distribution shifts, localizing the features originating them is still challenging, with current solutions being either inaccurate or not scalable to large and high-dimensional datasets. In this work, we introduce the Feature Shift Localization Network (FSL-Net), a neural network that can localize feature shifts in large and high-dimensional datasets in a fast and accurate manner. The network, trained with a large number of datasets, learns to extract the statistical properties of the datasets and can localize feature shifts from previously unseen datasets and shifts without the need for re-training. The code and ready-to-use trained model are available at https://github.com/AI-sandbox/FSL-Net.

Problem

• Localizing feature shifts in heterogeneous data sources
• Addressing inaccurate or non-scalable current solutions
• Handling large and high-dimensional datasets efficiently

Innovation

• Neural network localizes feature shifts
• Handles large high-dimensional datasets
• No re-training needed for new data