Bidirectional Generative Modeling Using Adversarial Gradient Estimation
arXiv:2002.09161
Abstract
This paper considers the general -divergence formulation of bidirectional generative modeling, which includes VAE and BiGAN as special cases. We present a new optimization method for this formulation, where the gradient is computed using an adversarially learned discriminator. In our framework, we show that different divergences induce similar algorithms in terms of gradient evaluation, except with different scaling. Therefore this paper gives a general recipe for a class of principled -divergence based generative modeling methods. Theoretical justifications and extensive empirical studies are provided to demonstrate the advantage of our approach over existing methods.
References in corpus (19)
- Auto-Encoding Variational Bayes
- Large Scale GAN Training for High Fidelity Natural Image Synthesis
- Self-Attention Generative Adversarial Networks
- Generative Modeling by Estimating Gradients of the Data Distribution
- Disentangling by Factorising
- Adversarially Learned Inference
- VEEGAN: Reducing Mode Collapse in GANs using Implicit Variational Learning
- Large Scale Adversarial Representation Learning
- Mode Regularized Generative Adversarial Networks
- Learning in Implicit Generative Models
- Adversarial Variational Bayes: Unifying Variational Autoencoders and Generative Adversarial Networks
- On Unifying Deep Generative Models
- Variational Inference using Implicit Distributions
- Gradient Estimators for Implicit Models
- Improved generator objectives for GANs
- A Spectral Approach to Gradient Estimation for Implicit Distributions
- On the Quantitative Analysis of Decoder-Based Generative Models
- Symmetric Variational Autoencoder and Connections to Adversarial Learning
- Mutual Information Gradient Estimation for Representation Learning