Bayesian Optimisation

Bayesian Optimization (BO) is a powerful technique for efficiently finding the optimal settings of a system or process whose behavior is represented by a "black-box" function, meaning its internal workings are unknown or too complex to model directly. Current research focuses on improving BO's scalability to high-dimensional problems and complex constraints, often employing Gaussian processes, neural processes, and novel acquisition functions to guide the search. These advancements are driving significant impact across diverse fields, including engineering design, materials science, and climate change mitigation, by enabling more efficient and effective optimization of complex systems with expensive or noisy evaluations.