FaLKM-lib v1.0: A Library for Fast Local Kernel Machines

FaLKM-lib v1.0 is a library for fast local kernel machine implemented in C++. It contains a fast implementation of kernel k-nearest neighbors (kNN) using the Cover Tree data-structure called FkNN, the local support vector machine (LSVM) algorithm called FkNNSVM, a noise reduction technique based on a probabilistic version of LSVM called FkNNSVM-nr, a fast and scalable version of LSVM called FaLK-SVM (subdivided in the two modules: FaLK-SVM-train and FaLK-SVM-predict) and a fast and scalable noise reduction technique called FaLKNR. The library contains tools for model selection, local model selection and automatic tuning of kernel parameters. This document introduces the formulation of the algorithms in the software library; for a comprehensive discussion on the implemented techniques please refer to the papers cited in this document and for the use of the software refer to the README file included in the package. FaLKM-lib v1.0 code is freely available for research and educational purposes at http://disi.unitn.it/~segata/FaLKM-lib.