Unsupervised Controllable Generation with Self-Training

by   Grigorios G. Chrysos, et al.

Recent generative adversarial networks (GANs) are able to generate impressive photo-realistic images. However, controllable generation with GANs remains a challenging research problem. Achieving controllable generation requires semantically interpretable and disentangled factors of variation. It is challenging to achieve this goal using simple fixed distributions such as Gaussian distribution. Instead, we propose an unsupervised framework to learn a distribution of latent codes that control the generator through self-training. Self-training provides an iterative feedback in the GAN training, from the discriminator to the generator, and progressively improves the proposal of the latent codes as training proceeds. The latent codes are sampled from a latent variable model that is learned in the feature space of the discriminator. We consider a normalized independent component analysis model and learn its parameters through tensor factorization of the higher-order moments. Our framework exhibits better disentanglement compared to other variants such as the variational autoencoder, and is able to discover semantically meaningful latent codes without any supervision. We demonstrate empirically on both cars and faces datasets that each group of elements in the learned code controls a mode of variation with a semantic meaning, e.g. pose or background change. We also demonstrate with quantitative metrics that our method generates better results compared to other approaches.


page 7

page 8


Intra-class Variation Isolation in Conditional GANs

Current state-of-the-art conditional generative adversarial networks (C-...

Unbalanced GANs: Pre-training the Generator of Generative Adversarial Network using Variational Autoencoder

We propose Unbalanced GANs, which pre-trains the generator of the genera...

Instance-Conditioned GAN

Generative Adversarial Networks (GANs) can generate near photo realistic...

Navigating the GAN Parameter Space for Semantic Image Editing

Generative Adversarial Networks (GANs) are currently an indispensable to...

InfoGAN-CR: Disentangling Generative Adversarial Networks with Contrastive Regularizers

Training disentangled representations with generative adversarial networ...

Semantically Decomposing the Latent Spaces of Generative Adversarial Networks

We propose a new algorithm for training generative adversarial networks ...

Contrastive Learning for Diverse Disentangled Foreground Generation

We introduce a new method for diverse foreground generation with explici...

Please sign up or login with your details

Forgot password? Click here to reset