 |
Software / RelEns - Relational Ensemble Classification
RelEns is a java tool for classifying instances by using the relations to other instances (i.e. the classes of neighbors) and combining results of different relations with ensemble classification methods. The instances could be time series or any other type.
RelEns has been funded by X-Media.
Developers: Christine Preisach
Download:
- Including jar file, the dtd for building the network of neighbors, a short documentation and javadoc (all compiled, all needed libraries included, ready to use) : RelEns.zip
- Results for the UCR benchmark datasets using 5 times 5-fold cross validation protocol: resultsUCRCV.pdf
Related publications:
- Christine Preisach, Lars Schmidt-Thieme (2006):
Relational Ensemble Classification, in Proceedings of the 6th IEEE International Conference on Data Mining (ICDM) 2006, Hong Kong.
- Christine Preisach, Lars Schmidt-Thieme (2008):
Ensembles of Relational Classifiers, Knowledge and Information Systems Journal 14(3).
System Requirements
- JDK >= 1.5