Incremental Uncertainty-aware Performance Monitoring with Active Labeling Intervention

TL;DR

This paper introduces IUPM, a novel method for monitoring machine learning model performance under gradual distribution shifts, using optimal transport and active labeling to maintain accuracy without extensive labeling.

Contribution

It presents a new label-free, uncertainty-aware performance monitoring approach that effectively tracks model accuracy during slow distribution changes.

Findings

IUPM outperforms existing baselines in gradual shift scenarios.

Uncertainty awareness improves label acquisition efficiency.

Method effectively maintains performance estimates with limited labels.

Abstract

We study the problem of monitoring machine learning models under gradual distribution shifts, where circumstances change slowly over time, often leading to unnoticed yet significant declines in accuracy. To address this, we propose Incremental Uncertainty-aware Performance Monitoring (IUPM), a novel label-free method that estimates performance changes by modeling gradual shifts using optimal transport. In addition, IUPM quantifies the uncertainty in the performance prediction and introduces an active labeling procedure to restore a reliable estimate under a limited labeling budget. Our experiments show that IUPM outperforms existing performance estimation baselines in various gradual shift scenarios and that its uncertainty awareness guides label acquisition more effectively compared to other strategies.