Extendable and invertible manifold learning with geometry regularized autoencoders | IEEE Conference Publication | IEEE Xplore