Simultaneous Learning and Reshaping of an Approximated Optimization Task

For many target optimization and learning tasks the sample cost of performing the task is very expensive or time consuming such that attempting to directly employ a learning algorithm on the task becomes intractable. For this reason learning is instead often performed on a less expensive task that is believed to be a reasonable approximation of the actual target task. This paper serves to present and motivate the challenging open problem of simultaneously performing learning on an approximation of the true target task, while at the same time shaping the task used for learning to be a better representation of the true target task. Our work, which is still in progress, is performed in the RoboCup 3D simulation environment where we attempt to learn walk parameters for an omnidirectional walk engine used by humanoid robot soccer playing agents.