Learning Controllable Disentangled Representations with Decorrelation Regularization
arXiv:1912.11675
Abstract
A crucial problem in learning disentangled image representations is controlling the degree of disentanglement during image editing, while preserving the identity of objects. In this work, we propose a simple yet effective model with the encoder-decoder architecture to address this challenge. To encourage disentanglement, we devise a distance covariance based decorrelation regularization. Further, for the reconstruction step, our model leverages a soft target representation combined with the latent image code. By exploiting the real-valued space of the soft target representations, we are able to synthesize novel images with the designated properties. We also design a classification based protocol to quantitatively evaluate the disentanglement strength of our model. Experimental results show that the proposed model competently disentangles factors of variation, and is able to manipulate face images to synthesize the desired attributes.
References in corpus (12)
- Auto-Encoding Variational Bayes
- Conditional Generative Adversarial Nets
- Stochastic Backpropagation and Approximate Inference in Deep Generative Models
- Improved Techniques for Training GANs
- InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
- Deep Convolutional Inverse Graphics Network
- Adversarially Learned Inference
- Invertible Conditional GANs for image editing
- Adversarial Autoencoders
- Unsupervised Learning of Disentangled Representations from Video
- Multi-Level Variational Autoencoder: Learning Disentangled Representations from Grouped Observations
- A Survey of Inductive Biases for Factorial Representation-Learning