A Generic Tool for Performance Evaluation of Supervised Learning Algorithms
Ron Begleiter and Ran El-Yaniv
Computer Science, Technion, Israel.
We provide a general purpose Matlab tool for performance evaluation of supervised learning algorithms. Given any existing implementation of a classifier learning algorithm, our Generic-CV tool operates the algorithm, tunes its hyper-parameters and evaluates its performance on any given dataset. The tool is generic in the sense that it easily accommodates new learning algorithms, without any need for
programming, and can operate Matlab implementations as well as `executables'. In addition, we provide a distributed version of our tool that can utilize multiple computers and in principle can speed up the enormous running time required for performance evaluation by several orders of magnitude.