Two Methods For Wild Variational Inference

2016/11/30 by Qiang Liu, Liu, Qiang, Yihao Feng +1 · 6 citations
Computer Science · Mathematics · Physics and Astronomy · #Gaussian Processes and Bayesian Inference #Markov Chains and Monte Carlo Methods #Model Reduction and Neural Networks #stat.ML

paper · pdf · doi:10.48550/arxiv.1612.00081

arxiv created 2017/10/30 · arxiv updated 2017/11/01

Abstract

Variational inference provides a powerful tool for approximate probabilistic in- ference on complex, structured models. Typical variational inference methods, however, require to use inference networks with computationally tractable proba- bility density functions. This largely limits the design and implementation of vari- ational inference methods. We consider wild variational inference methods that do not require tractable density functions on the inference networks, and hence can be applied in more challenging cases. As an example of application, we treat stochastic gradient Langevin dynamics (SGLD) as an inference network, and use our methods to automatically adjust the step sizes of SGLD, yielding significant improvement over the hand-designed step size schemes

Citations

• Auto-Encoding Variational Bayes

Cited by

• Stochastic Normalizing Flows
• Advances in Variational Inference
• Gradient Estimators for Implicit Models
• The reproducing Stein kernel approach for post-hoc corrected sampling
• Variational Implicit Processes
• Sliced Kernelized Stein Discrepancy

Related

• On the Convergence of Stochastic Variational Inference in Bayesian Networks
• VISA: Variational Inference with Sequential Sample-Average Approximations
• Langevin Diffusion Variational Inference
• Advances in Variational Inference
• Natural Evolution Strategies as a Black Box Estimator for Stochastic Variational Inference