An active learning method for solving competitive multi-agent decision-making and control problems