1 paper
Laurent Cheret, Vincent Létourneau, Isar Nejadgholi +3
We study a simple unsupervised regularization scheme for autoencoders called Manifold-Matching (MMAE): we align the pairwise distances in the latent space to those of the input dat…