Neural nets can be universal approximators for fuzzy functions | IEEE Conference Publication | IEEE Xplore