1 paper · 1 filter
Thomas Hunt, Arthur J. Krener
We present a new manifold learning algorithm that takes a set of data points lying on or near a lower dimensional manifold as input, possibly with noise, and outputs a simplicial c…