The name EGO is an acronym for efficient global optimization. The core concept of this method is to iteratively approximate the objective function using a Gaussian process surrogate model, which predicts both the objective function value and the associated uncertainty at un-sampled points. The Gaussian process is refined by sequentially selecting new sampling points based on an Expected Improvement acquisition function, which balances exploration (sampling where the model is uncertain) and exploitation (sampling where the model predicts improve objective values).