Learning to Draw Samples with Amortized Stein Variational Gradient Descent
This provides a method for flexible probabilistic inference in machine learning, though it appears incremental as it builds on existing Stein variational gradient descent techniques.
The paper tackles the problem of training stochastic neural networks to draw samples from target distributions for probabilistic inference, achieving this by adjusting network parameters along a Stein variational gradient direction to minimize KL divergence, and demonstrates applications in variational autoencoders and hyper-parameter learning for MCMC samplers.
We propose a simple algorithm to train stochastic neural networks to draw samples from given target distributions for probabilistic inference. Our method is based on iteratively adjusting the neural network parameters so that the output changes along a Stein variational gradient direction (Liu & Wang, 2016) that maximally decreases the KL divergence with the target distribution. Our method works for any target distribution specified by their unnormalized density function, and can train any black-box architectures that are differentiable in terms of the parameters we want to adapt. We demonstrate our method with a number of applications, including variational autoencoder (VAE) with expressive encoders to model complex latent space structures, and hyper-parameter learning of MCMC samplers that allows Bayesian inference to adaptively improve itself when seeing more data.