veroku 0.0.28

An open source library for building and performing inference with probabilistic graphical models.

Disclaimer

This project is still in the very early stages of development. Use this library at your own risk. The github repo is also still private, but will be made public soon.

Overview

Veroku is an open source library for building and performing inference with probabilistic graphical models (PGMs). PGMs provide a framework for performing efficient probabilistic inference with very high dimensional distributions. A typical example of a well-known type of PGM is the Kalman filter that can be used to obtain probabilistic estimates of a hidden state of a process or system, given noisy measurements. PGMs can in principle be used for any problem that involves uncertainty and is therefore applicable to many problems.

Veroku currently supports the following distributions:

• Multivariate Gaussian
• Multivariate Gaussian mixture
• Multivariate categorical

These distributions can be used as factors to represent a factorised distribution. These factors can be used, together with the `cluster_graph` module to automatically create valid cluster graphs. Inference can be performed in these graphs using message passing algorithms. Currently only the LBU (Loopy Belief Update) message-passing algorithm is supported.

Future Features

To be added soon:

• Example notebooks
• Non-linear Gaussian distribution
• Plate models (for efficiently specifying PGMs as modular/hierarchical templates)

On the roadmap:

• Dirichlet distribution
• Wishart distribution
• Normal-Wishart distribution