Correcting Sensor-Induced Distribution Drift with Wasserstein Adversarial Learning

2026-06-16

AI Summary

Sensor aging or displacement induces data distribution shifts that degrade the performance and stability of downstream data-driven models. This work proposes an unsupervised calibration method inspired by Wasserstein GANs, wherein the generator is explicitly modeled as a physically interpretable, differentiable calibration transformation. Through adversarial learning without labeled data, the method aligns the output distribution of the transformed sensor readings to a reference distribution, with the discriminator providing gradient signals based on the Wasserstein distance. By integrating unsupervised domain adaptation with high-granularity Geant4 simulation data, the approach successfully recovers aging coefficients for individual sensing units on simulated detectors—showing significant correlation with ground-truth values—and substantially improves the consistency between the calibrated energy response and the reference distribution.

Abstract

The quality of recorded data depends on the stability of the sensor system that acquires it. Sensor motion and aging can degrade the performance and stability of downstream data-driven methods. We present a Wasserstein-GAN-inspired approach for unsupervised inference of physically interpretable transformation parameters that map a changed detector response distribution back to a nominal reference distribution. In contrast to standard generative modeling, the generator is used as a learnable calibration transformation whose trainable weights represent the sought parameters, while the critic provides a distributional distance signal via the Wasserstein objective. We validate the approach on a tracking-detector toy model with controlled layer shifts and demonstrate its application on high-granularity Geant4-simulated calorimeter data with cell-wise aging effects. The method recovers aging coefficients for individual cells with correlation to ground truth and improves agreement between calibrated and reference energy-sum distributions, while exhibiting the expected degradation at increasing channel-to-channel noise levels. These results indicate that adversarial distribution matching can serve as a data-driven component of calibration strategies in settings where direct labels for degradation parameters are unavailable.

Problem

• distribution drift
• sensor aging
• calibration
• unsupervised learning
• Wasserstein distance

Innovation

• Wasserstein adversarial learning
• distribution drift correction
• unsupervised calibration
• sensor aging
• generative modeling