Percept is a modular machine learning framework that allows for easy testing and deployment.
A project is broken down into multiple tasks, which are run with a workflow. Workflows can be stored, manipulated, versioned, and accessed through an ipython shell. Tasks can be easily tested, chained together, given arguments, and used with dependencies. Different caching and data translation layers (ie, from the database to a pandas dataframe) can easily be added. Settings and configuration files allow for quick changes to be made. Percept tries to strike an optimal balance between development speed and deployability.
This is licensed under the Apache license, please see LICENSE.txt and NOTICE.txt for details. Alternate licensing is possible; contact email@example.com.
You can find documentation, including installation and usage guides, in the docs folder, or you can find the build version online at readthedocs.
Mailing List/IRC Channel
Feel free to open an issue in the issue tracker, or use this link to access an IRC chat room about this project. Alternatively, you can use your own IRC chat client and point it at freenode(chat.freenode.net). The channel is #equirio.
We also have a mailing list. Feel free to ask questions and discuss the code there. If you find an issue (unless it is a security issue, see below), please mention it on our issue tracker.
How to Contribute
Contributions are very welcome. The easiest way is to fork this repo, and then make a pull request from your fork.
The current backlog is in the issues section. Please feel free to open new issues or work on existing ones.
Reporting Security Issues
Please do not report security issues in public. Please email firstname.lastname@example.org.