This package implements the Informative Vector Machine algorithm discussed in the paper http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.2.4533&rep=rep1&type=pdf
This is a framework for Sparse Gaussian Process Regression. This algorithm returns the indexes of an active subset (d<<n) of datapoints which contains more information. This method can significantly increase the training speed for Gaussian Process regression.
The selection of datapoints is based on Information theory, where the predictors which minimizes the entropy of posterior process is greedily selected in an iterative fashion