Random Matrix Theory Proves that Deep Learning Representations of GAN-data Behave as Gaussian Mixtures

Abstract

This paper shows that deep learning (DL) representations of data produced by generative adversarial nets (GANs) are random vectors which fall within the class of so-called \textit{concentrated} random vectors. Further exploiting the fact that Gram matrices, of the type $G=X^{T}X$ with $X=[x_1,\dots ,x_n]\in {\mathbb{R}}^{p\times n}$ and $x_i$ independent concentrated random vectors from a mixture model, behave asymptotically (as $n,p\to \infty$) as if the $x_i$ were drawn from a Gaussian mixture, suggests that DL representations of GAN-data can be fully described by their first two statistical moments for a wide range of standard classifiers. Our theoretical findings are validated by generating images with the BigGAN model and across different popular deep representation networks.

References in corpus (9)

• Conditional Generative Adversarial Nets
• Spectral Normalization for Generative Adversarial Networks
• Improved Training of Wasserstein GANs
• Data Augmentation Generative Adversarial Networks
• Stabilizing Training of Generative Adversarial Networks through Regularization
• Representer Point Selection for Explaining Deep Neural Networks
• Concentration of Measure and Large Random Matrices with an application to Sample Covariance Matrices
• PCA of high dimensional random walks with comparison to neural network training
• A random matrix analysis and improvement of semi-supervised learning for large dimensional data

Cited by in corpus (8)

• Generalisation error in learning with random features and the hidden manifold model
• Learning curves of generic features maps for realistic datasets with a teacher-student model
• Universality Laws for High-Dimensional Learning with Random Features
• Concentration of Measure and Large Random Matrices with an application to Sample Covariance Matrices
• Statistical guarantees for generative models without domination
• Sparse Quantized Spectral Clustering
• On some theoretical limitations of Generative Adversarial Networks
• Resolvent convergence under heterogeneous second-moment profiles and quadratic-form control