Abstract: In this study, we explore the potential of machine learning models as surrogates for computationally intensive probabilistic models. Traditional multi-agent models (MAM) for forecasting ...