Welcome to Operalib’s documentation!¶
Operalib is a Python library devoted to multiple output regression and structured output prediction. It is mainly based on Operator-Valued Kernels (OVKs) and Output Kernels (OKs).
OVKs allow, for instance, to model vector-valued functions. In this senario it is possible to predict silmultaneously several targets while taking into account the dependence relationship between the coordinates of the output vector. First introduced in Machine-Learning to solve Multi-Task Regression problems, OVKs have been developed in the context of vector-valued Reproducing Kernel Hilbert Spaces (vv-RKHS) to address various tasks involving vectorial outputs: multiple structured classification, vector field learning, vector autoregression, functional regression, structured output prediction, link prediction…
Our aim is to provide an easy-to-use standard implementation of operator-valued kernel methods in the spirit of scikit-learn. Hence Operalib is designed for compatilibity to scikit-learn interface and conventions. It uses numpy, and scipy as underlying libraries.
The project is developed by:
LTCI, Télécom ParisTech,
IBISC, University of Evry, France.
See the README for additional information.