Fast learning of approximation policies for coordination in distributed networks | IEEE Conference Publication | IEEE Xplore