Active Learning Boosts Data-Driven Reduced Models for Digital Twins
Published:Jan 5, 2026 05:00
•1 min read
•ArXiv Stats ML
Analysis
This paper presents a valuable active learning framework for improving the efficiency and accuracy of reduced-order models (ROMs) used in digital twins. By intelligently selecting training parameters, the method enhances ROM stability and accuracy compared to random sampling, potentially reducing computational costs in complex simulations. The Bayesian operator inference approach provides a probabilistic framework for uncertainty quantification, which is crucial for reliable predictions.
Key Takeaways
Reference
“Since the quality of data-driven ROMs is sensitive to the quality of the limited training data, we seek to identify training parameters for which using the associated training data results in the best possible parametric ROM.”