Expert-Driven Monitoring of Operational ML Models
This addresses the challenge of maintaining ML model performance over time for practitioners in operational settings, but it appears incremental as it builds on existing monitoring approaches by incorporating expert knowledge.
The paper tackles the problem of concept drift in operational ML models by introducing Expert Monitoring, which leverages domain expertise to enhance detection and mitigation, supporting practitioners with consolidated expertise and automatic adaptability under expert oversight.
We propose Expert Monitoring, an approach that leverages domain expertise to enhance the detection and mitigation of concept drift in machine learning (ML) models. Our approach supports practitioners by consolidating domain expertise related to concept drift-inducing events, making this expertise accessible to on-call personnel, and enabling automatic adaptability with expert oversight.