Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Manifold-Matching Autoencoders (MMAE) are introduced, which regularize autoencoders by aligning pairwise distances in the latent space to those in the input data space using mean squared error. This distance-based alignment allows for extension to lower-dimensional representations. Empirically, MMAE outperforms existing methods on nearest-neighbor distance preservation and persistent homology metrics. MMAE also provides a scalable approximation of Multi-Dimensional Scaling (MDS).
Aligning pairwise distances in autoencoder latent spaces yields superior manifold preservation compared to existing regularization techniques, offering a scalable MDS approximation as a bonus.
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 data space by minimizing mean squared error. Because alignment occurs on pairwise distances rather than coordinates, it can also be extended to a lower-dimensional representation of the data, adding flexibility to the method. We find that this regularization outperforms similar methods on metrics based on preservation of nearest-neighbor distances and persistent homology-based measures. We also observe that MMAE provides a scalable approximation of Multi-Dimensional Scaling (MDS).