Comparison of Maximum Likelihood and GAN-based training of Real NVPs
read the original abstract
We train a generator by maximum likelihood and we also train the same generator architecture by Wasserstein GAN. We then compare the generated samples, exact log-probability densities and approximate Wasserstein distances. We show that an independent critic trained to approximate Wasserstein distance between the validation set and the generator distribution helps detect overfitting. Finally, we use ideas from the one-shot learning literature to develop a novel fast learning critic.
This paper has not been read by Pith yet.
Forward citations
Cited by 3 Pith papers
-
Guided Image Generation with Conditional Invertible Neural Networks
Proposes cINN architecture for conditional image generation that by construction yields diverse sharp samples, demonstrated on MNIST digit generation and image colorization with latent space manipulation.
-
Demystifying MMD GANs
MMD GANs have unbiased critic gradients but biased generator gradients from sample-based learning, and the Kernel Inception Distance provides a practical new measure for GAN convergence and dynamic learning rate adaptation.
-
Justifying Diagnosis Decisions by Deep Neural Networks
A multi-task deep learning model maps frontal X-rays to continuous text for producing diagnoses, textual justifications, and alternative images, with expert study showing better justification than saliency maps.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.