Mengjie Zhang, Evolving Loop and Recursion Structures in Genetic Programming
Evolving Loop and Recursion Structures in Genetic Programming
As a new paradigm, genetic programming has been successfully applied to many applications, from regression, circuit design, image processing and recognition, binary classification to recently multiclass classification problems. It is recognised as one of the main paradigm in machine learning and evolutionary computing.
Compared with human written programs, the genetic programs evolved by GP still have a number of limitations. While human written programs can have selection structure such as if-then, if-then-else and switch-case, loop structures such as for-loop, while-loop and do-while-loop, and other structures such as trees, lists and vectors, the GP evolved programs usually only have simple selection structure such as if-then and simple tree structures.
The goal of this project is to investigate new ways of introducing new structures into GP. We expect that this approach will extend the power of GP and improve the comprehensibility of genetic programs, and accordingly turn GP to a widely applicable technology for solving practical and more difficult problems. This project will focus on evolution of loop and recursion structure in genetic programming. Specifically, this project will investigate:
- Whether some high level of selection structures such as if-then-else and switch-case can be evolved in genetic programming;
- Whether loop structures such as for-loop, while-loop and do-while loop can be evolved in genetic programming; and
- Whether recursion structures can be evolved in genetic programming.
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).
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.