Welcome to cardinal’s documentation!

cardinal is a python package that allows simple and scikit-learn compatible active learning experiments. It provides the building blocks necessary to perform simple or more complex query sampling strategies.

The code of the project is on Github: cardinal

Installation

Using Pypi

cardinal can be installed through Pypi using:

$ pip install "cardinal[extra]"

Extras bring additional features to cardinal:

  • sklearn allows to use the KMeansSampler and RankedBatchModeSampler

  • examples intalls all packages necessary to run the examples

  • submodular installs apricot-select, and allows using SubmodularSampler

  • doc installs sphinx-gallery to be able to generate the documentation

  • all is an alias on all of the above.

Installing locally

You can also fetch the code and install the package from your local repository. Again, the preferred way is to use pip.

$ git clone https://github.com/dataiku-research/cardinal
$ cd cardinal
$ pip install -e ".[extra]"

Active learning

Example galleries

API and developer reference

Indices and tables