Learning optimal values from random walk | IEEE Conference Publication | IEEE Xplore