In this paper, a novel function approximation approach based on a combination of conventional Neural Networks and tile coding approximators is proposed. The proposed approach can maintain the desired features of both approaches whiles eliminates the deficiencies of each method. The combination will reduce the sharpness of tile coding. It will also provide an easy way to adjust the accuracy/complexity of the approximation according to the function being approximated (adaptive tiling) and the subspace used on. In this algorithm, it is possible to construct the approximator with specified and various approximation accuracies in different subspaces. This feature enables us to allocate an arbitrary accuracy/complexity wherever a more accurate approximation is needed. Finally simulation studies are presented to show the efficiency of and applicability of the proposed approach.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.