Skip to yearly menu bar Skip to main content


Invited 3
in
Workshop: Optimal Transport and Machine Learning

Optimal planar transport in near-linear time

Alexandr Andoni

[ ]
2017 Invited 3

Abstract:

We show how to compute the Earth Mover Distance between two planar sets of size N in N^{1+o(1)} time. The algorithm is based on a generic framework that decomposes the natural Linear Programming formulation for the transport problem into a tree of smaller LPs, and recomposes it in a divide-and-conquer fashion. The main enabling idea is use sketching -- a generalization of the dimension reduction method -- in order to reduce the size of the "partial computation" so that the conquer step is more efficient. We will conclude with some open questions in the area.

This is joint work with Aleksandar Nikolov, Krzysztof Onak, and Grigory Yaroslavtsev.

Live content is unavailable. Log in and register to view live content