Artificial Neural Networks as Surrogate Models in Black Box Optimization

This arXiv paper presents research on applying artificial neural networks as surrogate models in black-box optimization problems. Black-box optimization deals with objective functions that lack analytical descriptions, relying only on input-output pairs obtained through non-analytical means such as sensors or numerical simulations. The work demonstrates how neural network surrogate models can accelerate the optimization process by approximating the expensive objective function evaluations. The approach is applicable across various engineering fields that employ numerical simulation technologies.