Variational Inference with Normalizing Flows
Danilo Jimenez Rezende, Shakir Mohamed · arXiv (Cornell University) · 2015
The choice of approximate posterior distribution is one of the core problems in variational inference. Most applications of variational inference employ simple families of posterior approximations in order to allow for efficient inference, focusing on mean-field or other simple structured approximations. This restriction has a significant impact on the quality of inferences made using variational methods. We introduce a new approach for specifying flexible, arbitrarily complex and scalable approximate posterior distributions. Our approximations are distributions constructed through a normalizing flow, whereby a simple initial density is transformed into a more complex one by applying a sequence of invertible transformations until a desired level of complexity is attained. We use this view of normalizing flows to develop categories of finite and infinitesimal flows and provide a unified view of approaches for constructing rich posterior approximations. We demonstrate that the theoretical advantages of having posteriors that better match the true posterior, combined with the scalability of amortized variational approaches, provides a clear improvement in performance and applicability of variational inference.
Cited by 1,253 works. doi:10.48550/arxiv.1505.05770 Open-access PDF
Authors
References
- Book Review: Pattern Recognition and Machine Learning
- Adaptive Subgradient Methods for Online Learning and Stochastic Optimization
- Auto-Encoding Variational Bayes
- An Introduction to Variational Methods for Graphical Models
- Simple statistical gradient-following algorithms for connectionist reinforcement learning
- Pattern Recognition and Machine Learning (Information Science and Statistics)
- Deep AutoRegressive Networks
- Stochastic variational inference
- Semi-Supervised Learning with Deep Generative Models
- Bayesian Learning via Stochastic Gradient Langevin Dynamics