Manifold-Matching Autoencoders
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908893975674880 |
|---|---|
| author | Cheret, Laurent Létourneau, Vincent Nejadgholi, Isar Drummond, Chris Osman, Hussein Al Fraser, Maia |
| author_facet | Cheret, Laurent Létourneau, Vincent Nejadgholi, Isar Drummond, Chris Osman, Hussein Al Fraser, Maia |
| contents | 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). |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_16568 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Manifold-Matching Autoencoders Cheret, Laurent Létourneau, Vincent Nejadgholi, Isar Drummond, Chris Osman, Hussein Al Fraser, Maia Machine Learning Artificial Intelligence 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). |
| title | Manifold-Matching Autoencoders |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2603.16568 |