On sparsity, extremal structure, and monotonicity properties of Wasserstein and Gromov-Wasserstein optimal transport plans
New mathematical proof shows GW optimal transport plans can be supported on permutations under specific conditions.
Deep Dive
Researcher Titouan Vayer's arXiv paper analyzes Gromov-Wasserstein (GW) distance properties compared to standard optimal transport. The work proves that when the conditionally negative semi-definite property holds, GW optimal plans become sparse and can be supported on permutations. This provides mathematical guarantees about the structure of transport plans used in machine learning for comparing datasets with different geometries, like graphs or shapes.
Why It Matters
Enables more efficient and interpretable geometric data matching in ML applications like graph alignment and shape analysis.