Operalib is a library for structured learning and prediction for python based on operator-valued kernels (OVKs). OVKs are an extension of scalar kernels to matrix-valued kernels. The idea is to predict silmultaneously several targets while, for instance, encoding the output structure with the operator-valued kernel.

We aim at providing an easy-to-use standard implementation of operator-valued kernel methods. Operalib is designed for compatilibity to scikit-learn interface and conventions. It uses numpy, scipy and cvxopt as underlying libraries.

The project is developed by the AROBAS group of the IBISC laboratory of the University of Evry, France.

Is available at: http://operalib.github.io/operalib/documentation/.

The package is available on PyPi, and the installation should be as simple as:

pip install operalib

This package uses distutils, which is the default way of installing python modules. To install in your home directory, use:

python setup.py install --user

To install for all users on Unix/Linux:

python setup.py build
sudo python setup.py install

You can check the latest sources with the command:

git clone https://github.com/operalib/operalib

or through ssh, instead of https, if you have write privileges:

git clone git@github.com:operalib/operalib.git

A non-exhaustive list of publications related to operator-valued kernel is available here:

http://operalib.github.io/operalib/documentation/reference_papers/index.html.