Neural Modes: Self-supervised Learning of Nonlinear Modal Subspaces
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911856720871424 |
|---|---|
| author | Wang, Jiahong Du, Yinwei Coros, Stelian Thomaszewski, Bernhard |
| author_facet | Wang, Jiahong Du, Yinwei Coros, Stelian Thomaszewski, Bernhard |
| contents | We propose a self-supervised approach for learning physics-based subspaces for real-time simulation. Existing learning-based methods construct subspaces by approximating pre-defined simulation data in a purely geometric way. However, this approach tends to produce high-energy configurations, leads to entangled latent space dimensions, and generalizes poorly beyond the training set. To overcome these limitations, we propose a self-supervised approach that directly minimizes the system's mechanical energy during training. We show that our method leads to learned subspaces that reflect physical equilibrium constraints, resolve overfitting issues of previous methods, and offer interpretable latent space parameters. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_17620 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Neural Modes: Self-supervised Learning of Nonlinear Modal Subspaces Wang, Jiahong Du, Yinwei Coros, Stelian Thomaszewski, Bernhard Machine Learning Computer Vision and Pattern Recognition Graphics We propose a self-supervised approach for learning physics-based subspaces for real-time simulation. Existing learning-based methods construct subspaces by approximating pre-defined simulation data in a purely geometric way. However, this approach tends to produce high-energy configurations, leads to entangled latent space dimensions, and generalizes poorly beyond the training set. To overcome these limitations, we propose a self-supervised approach that directly minimizes the system's mechanical energy during training. We show that our method leads to learned subspaces that reflect physical equilibrium constraints, resolve overfitting issues of previous methods, and offer interpretable latent space parameters. |
| title | Neural Modes: Self-supervised Learning of Nonlinear Modal Subspaces |
| topic | Machine Learning Computer Vision and Pattern Recognition Graphics |
| url | https://arxiv.org/abs/2404.17620 |