Keanu Docs

Keanu is a general-purpose probabilistic programming library. The main design goal is to build a Bayesian inference tool that scales to a point where we can model the world and therefore optimize it.

View on GitHub Get started

Installation

Learn how to install Keanu in a new or existing project.

Read more

What's it used for?

Learn what sort of problems Keanu can solve.

Read more

Getting started

Learn how to describe a model in Keanu.

Read more

How do I import and export my data?

Learn how to ingest and write out data to/from CSV.

Read more

What are the building blocks?

Learn how to use vertices - the fundamental components of models in Keanu.

Read more

How can I create vectors, matrices and higher dimensional arrays?

Learn about how tensors can be used to simplify your Bayesian network.

Read more

How do I calculate the most probable values of random variables in my model?

Learn how to use inference to find the most probable values for random variables given your data.

Read more

How do I calculate probabilities of events?

Learn how to infer the probability of some event given your data.

Read more

How do I create simple regression models?

Learn how to quickly construct linear/logistic regression models.

Read more

How can I infer different probable states of my network?

Learn how to use a Particle Filter to find probable states for your network.

Read more

How do I repeat logic in my model?

Learn how to use sequence templates to describe repeating subgraphs in Keanu or to do time series modelling.

Read more

Autocorrelation

Learn how to calculate autocorrelation on your samples.

Read more

Saving and Loading Models

Learn how to Save and Load constructed & trained models

Read more

Tutorial - Two thermometers

Learn from a real-world example.

Read more

Tutorial - Lorenz

Learn how to do inference even on chaotic systems

Read more