Surrogate Modelling

Surrogate modeling aims to replace computationally expensive simulations with faster, approximate models, enabling efficient exploration of complex systems in applications like design optimization and uncertainty quantification. Current research emphasizes the development and application of advanced machine learning techniques, including Bayesian neural networks, Gaussian processes, and deep operator networks, often incorporating domain knowledge or physics-informed constraints to improve accuracy and efficiency. This field is crucial for accelerating scientific discovery and engineering design across diverse domains, from materials science and fluid dynamics to cyber-physical systems and optimization problems, by enabling high-throughput simulations and robust uncertainty analysis.