Description

Rseslib is a library of rough set, machine learning and data mining data structures, algorithms and tools implemented in Java. The library was started by Group of Logic at Faculty of Mathematics, Informatics and Mechanics, University of Warsaw.

This web site introduces to the newest version of the library: Rseslib 3. The first version of the library was implemented in C++. Rseslib 2 was the first version implemented in Java and it stands for the core of RSES 2.x. Rseslib 3 was entirely redesigned and all the methods available in this version were implemented from scratch, 30+ people have already contributed to its source code and it is still developing.

Rseslib 3 provides:

  • wide range of open source algorithms and computational models from rough set theory, machine learning and data mining in Java
  • modular component-based architecture where source code components are easily reused and shared among algorithms
  • QMAK, Simple Grid Manager, port to Weka and command-line programs

Rseslib classifiers have been proven by independent researchers to rank among the classifiers with the highest classification accuracy (see Rseslib achievements).

Tools

Weka

Rseslib is available in Weka as official Weka package. The instruction on how to install Rseslib in Weka can be found in Chapter 14 WEKA of Rseslib User Guide.

QMAK

Graphical tool allowing users to interact with machine learning models and visualizing classification process.

Simple Grid Manager

Client-server tool for running Rseslib experiments on many computers or cores. Tutorial on how to run computations in cluster can be found in Chapter 16 SGM: Computing many experiments on many computers/cores of Rseslib User Guide.

Projects using Rseslib

  • TunedIT - system for automated evaluation, benchmarking and comparison of data mining and machine learning algorithms
  • Debellor - framework for scalable data mining and machine learning with data streaming
  • mahout-extensions - attribute selection extensions to Mahout, an extensible programming environment and framework for building scalable algorithms in machine learning
  • DMEXL - data mining expression library facilitating development of concurrent data mining algorithms