OKECA is a kernel feature extraction method based on entropy estimation in Hilbert spaces. It provides sparse and compact results, useful for data visualization and dimensionality reduction.