Active learning Active learning
Control Dynamics and control
FSI Fluid–structure interaction
Acoustics Acoustics and vibrations

Bayesian active-learning algorithms are widely used for design and optimization of black-box functions, largely because of their ability to compromise between exploration and exploitation before each function query. However, the criteria commonly used for sample selection fall short when the black box has the ability to generate rare and extreme events, i.e., events that combine high-magnitude impact with low probability of occurrence. We develop new criteria that guide the algorithm towards highly anomalous—and therefore highly relevant—regions of the search space. Applications include uncertainty quantification, optimization, extreme-event prediction, and environment monitoring.

Read more
Path planning for anomaly detection: Ocean Engineering 243, 110242 (2022)
Rare-event quantification: SIAM Journal on Uncertainty Quantification 9, 564–592 (2021)
Bayesian optimization: Journal of Computational Physics 425, 109901 (2021)