Back to Search
Start Over
Analyzing the latent space of GAN through local dimension estimation for disentanglement evaluation.
- Source :
-
Pattern Recognition . Jan2025, Vol. 157, pN.PAG-N.PAG. 1p. - Publication Year :
- 2025
-
Abstract
- The impressive success of style-based GANs (StyleGANs) in high-fidelity image synthesis has motivated research to understand the semantic properties of their latent spaces. In this paper, we approach this problem through a geometric analysis of latent spaces as a manifold. In particular, we propose a local dimension estimation algorithm for arbitrary intermediate layers in a pre-trained GAN model. The estimated local dimension is interpreted as the number of possible semantic variations from this latent variable. Moreover, this intrinsic dimension estimation enables unsupervised evaluation of disentanglement for a latent space. Our proposed metric, called Distortion , measures an inconsistency of intrinsic tangent space on the learned latent space. Distortion is purely geometric and does not require any additional attribute information. Nevertheless, Distortion shows a high correlation with the global-basis-compatibility and supervised disentanglement score. Our work is the first step towards selecting the most disentangled latent space among various latent spaces in a GAN without attribute labels. • We propose a scheme to estimate local intrinsic dimensions in GAN latent spaces. • Our local dimension estimate provides an upper bound on local semantic perturbations. • We propose a layer-wise unsupervised disentanglement score, called Distortion. • We analyze the layers of the mapping network in StyleGANs through Distortion metric. [ABSTRACT FROM AUTHOR]
- Subjects :
- *GENERATIVE adversarial networks
*GEOMETRIC analysis
*FACTORIZATION
*ALGORITHMS
Subjects
Details
- Language :
- English
- ISSN :
- 00313203
- Volume :
- 157
- Database :
- Academic Search Index
- Journal :
- Pattern Recognition
- Publication Type :
- Academic Journal
- Accession number :
- 179603412
- Full Text :
- https://doi.org/10.1016/j.patcog.2024.110914