Strongly-consistent nonparametric estimation of smooth regression functions for stationary ergodic sequences

Let {(X/sub i/,Y/sub i/)} be a stationary ergodic R/sup d//spl times/R valued process. This study offers a strongly consistent (with respect to pointwise, least-squares, and uniform distance) algorithm for inferring the regression function E[Y/sub 0/|X/sub 0/=x], assumed uniformly Lipschitz continuous.