Hyperplane Arrangements and Fixed Points in Iterated PWL Neural Networks
Fuente:
arXiv
Enregistré dans:
| Auteur principal: | |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917721347719168 |
|---|---|
| author | Beise, Hans-Peter |
| author_facet | Beise, Hans-Peter |
| contents | We leverage the framework of hyperplane arrangements to analyze potential regions of (stable) fixed points. We provide an upper bound on the number of fixed points for multi-layer neural networks equipped with piecewise linear (PWL) activation functions with arbitrary many linear pieces. The theoretical optimality of the exponential growth in the number of layers of the latter bound is shown. Specifically, we also derive a sharper upper bound on the number of stable fixed points for one-hidden-layer networks with hard tanh activation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_09878 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Hyperplane Arrangements and Fixed Points in Iterated PWL Neural Networks Beise, Hans-Peter Machine Learning Artificial Intelligence 68T07 G.0 We leverage the framework of hyperplane arrangements to analyze potential regions of (stable) fixed points. We provide an upper bound on the number of fixed points for multi-layer neural networks equipped with piecewise linear (PWL) activation functions with arbitrary many linear pieces. The theoretical optimality of the exponential growth in the number of layers of the latter bound is shown. Specifically, we also derive a sharper upper bound on the number of stable fixed points for one-hidden-layer networks with hard tanh activation. |
| title | Hyperplane Arrangements and Fixed Points in Iterated PWL Neural Networks |
| topic | Machine Learning Artificial Intelligence 68T07 G.0 |
| url | https://arxiv.org/abs/2405.09878 |