Exploring Semantic Variations in GAN Latent Spaces via Matrix Factorization
This work addresses the problem of entangled manipulations in GAN latent spaces for researchers and practitioners, but it is incremental as it builds on existing methods like GANSpace.
The paper tackled the challenge of controlling data generation in GANs by exploring image manipulations in latent spaces, finding that replacing PCA with ICA improves disentanglement and quality, with fundamental controlling directions observed across GAN sizes.
Controlled data generation with GANs is desirable but challenging due to the nonlinearity and high dimensionality of their latent spaces. In this work, we explore image manipulations learned by GANSpace, a state-of-the-art method based on PCA. Through quantitative and qualitative assessments we show: (a) GANSpace produces a wide range of high-quality image manipulations, but they can be highly entangled, limiting potential use cases; (b) Replacing PCA with ICA improves the quality and disentanglement of manipulations; (c) The quality of the generated images can be sensitive to the size of GANs, but regardless of their complexity, fundamental controlling directions can be observed in their latent spaces.