Abstract: This paper presents a model-free neural network controller design methodology based on transfer reinforcement learning (TRL) with Gaussian reward shaping, implemented and validated on a Buck ...