Jesse Read

River is a machine learning library for dynamic data streams and continual learning. It provides multiple state-of-the-art learning methods, data generators/transformers, performance metrics and evaluators for different stream learning problems. It is the result from the merger of two popular packages for stream learning in Python: Creme and scikit-multiflow.

Scikit-Multiflow provides a Python framework for comparing, evaluating, and visualizing traditional and multi-output machine learning methods in a data-stream setting. Many methods from MEKA are reimplemented here (earlier, as part of the Molearn project). The Scikit-Learn framework supplies the ‘base models’.

MEKA (Multi-label/multi-target Extension to weKA) is a Java framework containing a few dozen multi-label and multi-output classifiers, and includes a graphical user interface. It includes methods and an evaluation framework incremental and semi-supervised learning, among other contexts.

I am involved in the MOA Project for learning in data streams.

I created ALife: an artificial life environment for testing and comparing reinforcement learning algorithms with continous action and state space.

A2RMS is a Matlab/GNU Octave implementation of the A2RMS sampling algorithm.

Code for implementing distributed particle filters, designed for use on very low powered sensor motes, in the TinyOS operating system (here is a video of a deployment with a sensor testbed).

However, the code is also standalone and can be compiled and run as a regular C library.