Mengjie Zhang, Genetic Programming for Feature Selection and Construction

Evolutionary Feature Selection and Big Dimensionary Reduction


Many real-world problems involve a large number of features/attributes/variable. Since the relative importance of them is usually unknown in advance, all of them are typically used to perform classification, regression or optimisation tasks. However, not all the features are critical fora particular task, the features are seldom equally important, and some features are too abstract to contribute to that task, which often lead to unsatisfactory performance.

Due to the powerful search abilities and/or flexible solution encoding/representation schemes, evolutionary computation (EC) techniques have shown great potential to solve these problems. However, the dimensionality and the complexity of the data in real-world problems grows fast in recent years, which requires novel effective and efficient approaches to addressing new challenges in this area.

The goal of this project is to investigate novel feature discovery and construction algorithms to deal with the big dimensionality issue in classification, regression and optimisation tasks. Specifically, at least one of the following research topics will be considered in the project:

  • Feature ranking/weighting, subset selection and construction;
  • Novel fitness evaluation criteria in feature selection and construction;
  • Feature extraction/construction in image data;
  • Single objective and multi-objective feature selection and construction;
  • Scalability in feature selection and construction for big dimensionality;
  • Biomarker Discovery in Proteomics Mass Spectrometry (Bioinformatics)

A strong background in Java/C/C++ programming and a basic background in Artificial Intelligence and statistics are required. A good background in machine learning, and operations research is desired (COMP307, COMP361).

This project will be co-supervised by Dr Bing Xue. The School has good international reputation in the field and would like to continue the momentum. Please check http://homepages.ecs.vuw.ac.nz/~mengjie/papers/index.shtml, http://ecs.victoria.ac.nz/Main/MengjieZhang, and http://ecs.victoria.ac.nz/Groups/ECRG/ for publications and other information.