Variational Message Passing

Non-conjugate Variational Message Passing for Multinomial and Binary Regression

Variational Message Passing (VMP) is an algorithmic implementation of the Variational Bayes (VB) method which applies only in the special case of conjugate exponential family models. We propose an extension to VMP, which we refer to as Non-conjugate Variational Message Passing (NCVMP) which aims to alleviate this restriction while maintaining modularity, allowing choice in how expectations are ...

Centralized Estimation of Adhesion Loss in Wheel – Rail System Using Variational Bayes and Variational Message Passing

Gates

Gates are a new notation for representing mixture models and context-sensitive independence in factor graphs. Factor graphs provide a natural representation for message-passing algorithms, such as expectation propagation. However, message passing in mixture models is not well captured by factor graphs unless the entire mixture is represented by one factor, because the message equations have a c...

Gates: A graphical notation for mixture models

Gates are a new notation for representing mixture models and context-sensitive independence in factor graphs. Factor graphs provide a natural representation for message-passing algorithms, such as expectation propagation. However, message passing in mixture models is not well captured by factor graphs unless the entire mixture is represented by one factor, because the message equations have a c...

Sparse Message Passing and Efficiently Learning Random Fields for Stereo Vision

Message passing algorithms based on variational methods and belief propagation are widely used for approximate inference in a variety of directed and undirected graphical models. However, inference can become extremely slow when the cardinality of the state space of individual variables is high. In this paper we explore sparse message passing to dramatically accelerate approximate inference. We...

Fully simplified multivariate normal updates in non-conjugate variational message passing

Fully simplified expressions for Multivariate Normal updates in non-conjugate variational message passing approximate inference schemes are obtained. The simplicity of these expressions means that the updates can be achieved very efficiently. Since the Multivariate Normal family is the most common for approximating the joint posterior density function of a continuous parameter vector, these ful...