Composing graphical models with neural networks for structured representations and fast inference
This work addresses the challenge of integrating graphical models and neural networks for researchers in machine learning, offering a general framework that is incremental in nature.
The authors tackled the problem of combining probabilistic graphical models with deep learning to create structured representations and enable fast inference, resulting in a scalable algorithm that integrates stochastic variational inference, natural gradients, message passing, and the reparameterization trick, as demonstrated in an application to mouse behavioral phenotyping.
We propose a general modeling and inference framework that composes probabilistic graphical models with deep learning methods and combines their respective strengths. Our model family augments graphical structure in latent variables with neural network observation models. For inference, we extend variational autoencoders to use graphical model approximating distributions with recognition networks that output conjugate potentials. All components of these models are learned simultaneously with a single objective, giving a scalable algorithm that leverages stochastic variational inference, natural gradients, graphical model message passing, and the reparameterization trick. We illustrate this framework with several example models and an application to mouse behavioral phenotyping.