Software framework for modular machine learning systems

Abstract

Machine learning methods and algorithms can be combined into ensembles to obtain better performance than a single base learner. In the paper we present a framework for distributed system based on Common Object Request Broker Architecture for creating ensembles of learning systems. The systems are handled by the server which sends and receives learning and testing data. They can be located on different machines with various operating systems or hardware. The structures of the base learners are described by XML files.