Parallelized particle advection algorithms are a key visualization tool for domain scientists. They are also very computationally expensive to run. Machine learning techniques have been widely used in regression settings to predict results based on a set of input features. Our work describes an approach for parallel particle advection optimization; an approach which uses a machine learning algorithm at its core. We specifically investigate how our approach operates when applied to a GPU-based parallel particle advection algorithm. We examine the efficacy of 14 different machine learning models (including six neural network models) in our approach and validate the top theoretical performers' accuracy with respect to speedup over a baseline algorithm. From our investigations we find that the machine learning based optimization architecture yielded by our approach considers 45% of examined particle advection workloads to be "acceleratable," of which 74% truly do receive acceleration. Of the "acceleratable" workloads, our architecture achieves an overall average speedup of 12.7% and an average speedup of 20.3% for workloads that took longer than 60 seconds to execute.