Wasserstein GANs are Minimax Optimal Distribution Estimators
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915193006587904 |
|---|---|
| author | Stéphanovitch, Arthur Aamari, Eddie Levrard, Clément |
| author_facet | Stéphanovitch, Arthur Aamari, Eddie Levrard, Clément |
| contents | We provide non asymptotic rates of convergence of the Wasserstein Generative Adversarial networks (WGAN) estimator. We build neural networks classes representing the generators and discriminators which yield a GAN that achieves the minimax optimal rate for estimating a certain probability measure $μ$ with support in $\mathbb{R}^p$. The probability $μ$ is considered to be the push forward of the Lebesgue measure on the $d$-dimensional torus $\mathbb{T}^d$ by a map $g^\star:\mathbb{T}^d\rightarrow \mathbb{R}^p$ of smoothness $β+1$. Measuring the error with the $γ$-Hölder Integral Probability Metric (IPM), we obtain up to logarithmic factors, the minimax optimal rate $O(n^{-\frac{β+γ}{2β+d}}\vee n^{-\frac{1}{2}})$ where $n$ is the sample size, $β$ determines the smoothness of the target measure $μ$, $γ$ is the smoothness of the IPM ($γ=1$ is the Wasserstein case) and $d\leq p$ is the intrinsic dimension of $μ$. In the process, we derive a sharp interpolation inequality between Hölder IPMs. This novel result of theory of functions spaces generalizes classical interpolation inequalities to the case where the measures involved have densities on different manifolds. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2311_18613 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Wasserstein GANs are Minimax Optimal Distribution Estimators Stéphanovitch, Arthur Aamari, Eddie Levrard, Clément Statistics Theory We provide non asymptotic rates of convergence of the Wasserstein Generative Adversarial networks (WGAN) estimator. We build neural networks classes representing the generators and discriminators which yield a GAN that achieves the minimax optimal rate for estimating a certain probability measure $μ$ with support in $\mathbb{R}^p$. The probability $μ$ is considered to be the push forward of the Lebesgue measure on the $d$-dimensional torus $\mathbb{T}^d$ by a map $g^\star:\mathbb{T}^d\rightarrow \mathbb{R}^p$ of smoothness $β+1$. Measuring the error with the $γ$-Hölder Integral Probability Metric (IPM), we obtain up to logarithmic factors, the minimax optimal rate $O(n^{-\frac{β+γ}{2β+d}}\vee n^{-\frac{1}{2}})$ where $n$ is the sample size, $β$ determines the smoothness of the target measure $μ$, $γ$ is the smoothness of the IPM ($γ=1$ is the Wasserstein case) and $d\leq p$ is the intrinsic dimension of $μ$. In the process, we derive a sharp interpolation inequality between Hölder IPMs. This novel result of theory of functions spaces generalizes classical interpolation inequalities to the case where the measures involved have densities on different manifolds. |
| title | Wasserstein GANs are Minimax Optimal Distribution Estimators |
| topic | Statistics Theory |
| url | https://arxiv.org/abs/2311.18613 |