This repository contains Subutai's nupic code and experiments. Some of the directories are quite old and may not work with the latest versions of NuPIC. Nevertheless, they worked at one point, and may spark some ideas.
The directory swarm_experiments contain a tutorial demonstrating
swarming, handling multiple fields, noisy data, etc.
The directory run_anomaly contain some sample scripts for running
anomaly detection on numerical datasets.
The directory ensembles contains simple sample code for a predictor
that is an ensemble of HTM models. The ensemble gets better results
than any single HTM model.
The directory anomaly_classification contains a simple example showing
how to use classification for detecting known anomalies even if they
occur again.
The directory rns contains simple code for computing with residue
number systems. I did this mostly to build up some intuition for grid
cells.
I am a Numenta employee. As such this code is offered under the same terms as NuPIC. The license is AGPL and Numenta owns the original copyright. If you have a pull request or otherwise want to contribute to this repository, please ensure you have agreed to the Numenta Contributor's License Agreement first.