Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2411.03611 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913572685086720 |
|---|---|
| author | Akiyama, Keito |
| author_facet | Akiyama, Keito |
| contents | In recent years, learning for neural networks can be viewed as optimization in the space of probability measures. To obtain the exponential convergence to the optimizer, the regularizing term based on Shannon entropy plays an important role. Even though an entropy function heavily affects convergence results, there is almost no result on its generalization, because of the following two technical difficulties: one is the lack of sufficient condition for generalized logarithmic Sobolev inequality, and the other is the distributional dependence of the potential function within the gradient flow equation. In this paper, we establish a framework that utilizes a linearized potential function via Csiszár type of Tsallis entropy, which is one of the generalized entropies. We also show that our new framework enable us to derive an exponential convergence result. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_03611 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Designing a Linearized Potential Function in Neural Network Optimization Using Csiszár Type of Tsallis Entropy Akiyama, Keito Machine Learning Analysis of PDEs 35Q49, 49J20, 82C32 In recent years, learning for neural networks can be viewed as optimization in the space of probability measures. To obtain the exponential convergence to the optimizer, the regularizing term based on Shannon entropy plays an important role. Even though an entropy function heavily affects convergence results, there is almost no result on its generalization, because of the following two technical difficulties: one is the lack of sufficient condition for generalized logarithmic Sobolev inequality, and the other is the distributional dependence of the potential function within the gradient flow equation. In this paper, we establish a framework that utilizes a linearized potential function via Csiszár type of Tsallis entropy, which is one of the generalized entropies. We also show that our new framework enable us to derive an exponential convergence result. |
| title | Designing a Linearized Potential Function in Neural Network Optimization Using Csiszár Type of Tsallis Entropy |
| topic | Machine Learning Analysis of PDEs 35Q49, 49J20, 82C32 |
| url | https://arxiv.org/abs/2411.03611 |