RSSalg software: A tool for flexible experimenting with co-training based semi-supervised algorithms

Abstract

© 2017 RSSalg software is a tool for experimenting with Semi-Supervised Learning (SSL), a set of machine learning techniques able to use both labeled and unlabeled data for training. The goal is to reduce human effort regarding data labeling while preserving model quality. RSSalg software encompasses the implementation of co-training, a multi-view SSL technique and RSSalg, its single-view alternative. Our tool enables easy comparison of different SSL algorithms. It provides a cross-validation procedure and supports standard metrics for performance evaluation. The tool is free and open source, available on GitHub under the GNU General Public License. It is implemented in Java language using Weka library.