NeurIPS 2020

Adaptive Probing Policies for Shortest Path Routing

Meta Review

This paper gives an adaptive policy for choosing the shortest path between two given nodes in a network when the designer knows the distribution of the lengths but each edge-probe is expensive. The algorithm is based on a neat reduction to adaptive submodular optimization framework. The reviewers had a substantial discussion on the merits and assumptions of this work and were convinced that it gives a neat result on a elegant and important problem. I am pleased to recommend acceptance.