An Iterative Split Adjustment Multi-label Decision Tree

ISA-MDT is a method for constructing a decision tree when items in the dataset have multiple, and possibly weighted, labels. The algorithm was developed during my Ph.D. research.

Currently written in Matlab. A non-matlab, executable implementation to be written and made available at some point.


  1. Aiyesha Ma. “An Induction Method for Multilabel Decision Trees with Application to Multimedia Tagging and Other Problems”. Doctor of Philosophy in Systems Engineering. PhD thesis. Rochester, MI: Oakland University, 2010. (pdf)

  2. Aiyesha Ma, Ishwar K. Sethi, and Nilesh Patel. “Multi-label Classification Method for Multimedia Tagging”. In: International Journal of Multimedia Data Engineering and Management (IJMDEM) 1. 3 (2010), pp. 57–75. DOI: 10.4018/jmdem.2010070104.

  3. Aiyesha Ma and Ishwar K. Sethi. “Iterative Split Adjustment for Building Multilabel Decision Trees”. In: Proceedings of the ISCA 2nd International Conference on Bioinformatics and Computational Biology, BICoB-2010. (Sheraton Waikiki Hotel, Honolulu, Hawaii, USA, Mar. 24–26, 2010). Ed. by Hisham Al-Mubaid. ISCA, 2010, pp. 13–18. ISBN: 978-1-880843-76-5.

  4. Aiyesha Ma, Ishwar K. Sethi, and Nilesh Patel. “Multimedia Content Tagging using Multilabel Decision Tree”. In: Proceedings of the 5th IEEE International Workshop on Multimedia Information Processing and Retrieval. (San Diego, California, USA, Dec. 14–16, 2009). IEEE Computer Society, 2009, pp. 606–611. ISBN: 978-1-4244-5231-6. DOI: 10.1109/ISM.2009.87.