Vaughn, Ryan

Diffusion maps and other graph Laplacian based methods in machine learning have a long history of pointwise consistency results. This work provides rigorous formulation for diffusion maps for manifolds with boundary. We show that different normalizations of graph Laplacians are …