A Spectral Regularizer for Unsupervised Disentanglement
This addresses the challenge of unsupervised disentanglement for generative models, offering a method to control output aspects independently, though it appears incremental as it builds on existing GAN frameworks.
The paper tackles the problem of learning disentangled representations in GANs by showing that curved trajectories in latent space can exhibit local disentanglement, and it introduces a spectral regularizer that aligns these trajectories with coordinate axes to induce disentanglement in an unsupervised way.
A generative model with a disentangled representation allows for independent control over different aspects of the output. Learning disentangled representations has been a recent topic of great interest, but it remains poorly understood. We show that even for GANs that do not possess disentangled representations, one can find curved trajectories in latent space over which local disentanglement occurs. These trajectories are found by iteratively following the leading right-singular vectors of the Jacobian of the generator with respect to its input. Based on this insight, we describe an efficient regularizer that aligns these vectors with the coordinate axes, and show that it can be used to induce disentangled representations in GANs, in a completely unsupervised manner.