Reliably detecting model failures in deployment without labels
By: Viet Nguyen , Changjian Shui , Vijay Giri and more
Potential Business Impact:
Alerts when computer brains need to learn again.
The distribution of data changes over time; models operating operating in dynamic environments need retraining. But knowing when to retrain, without access to labels, is an open challenge since some, but not all shifts degrade model performance. This paper formalizes and addresses the problem of post-deployment deterioration (PDD) monitoring. We propose D3M, a practical and efficient monitoring algorithm based on the disagreement of predictive models, achieving low false positive rates under non-deteriorating shifts and provides sample complexity bounds for high true positive rates under deteriorating shifts. Empirical results on both standard benchmark and a real-world large-scale internal medicine dataset demonstrate the effectiveness of the framework and highlight its viability as an alert mechanism for high-stakes machine learning pipelines.
Similar Papers
Interpretable Model Drift Detection
Machine Learning (CS)
Finds when computer learning gets old.
Probabilistic Runtime Verification, Evaluation and Risk Assessment of Visual Deep Learning Systems
Machine Learning (CS)
Makes AI more trustworthy in the real world.
Incremental Uncertainty-aware Performance Monitoring with Active Labeling Intervention
Machine Learning (CS)
Finds when computer smarts are getting worse.