School of ITEE, UNSW@ADFA

Topic: Developmental Genetic Programming

This topic aims to investigate developmental Genetic Programming. Hierarchical structure has not so far been observed in extant developmental GP systems. Possibly this is because modular structure only has adaptive advantages to an overall species, and not to a particular individual, and hence cannot be selected for by evolution. The working hypothesis is that if an individual is evaluated multiple times throughout its development, then modular structure can provide an adaptive advantage to that individual, and hence can be selected for by evolution. The research plan is to investigate this hypothesis using the TAG3P; GP system developed in UNSW@ADFA, which has ideal properties for supporting evaluation during incremental development. Our preliminary indications on this are extremely promising, suggesting that positive results are highly likely. After confirming that developmental evaluation promotes the evolution of modular structure, we will then build a full-fledged developmental system, again based on the TAG representation, to take full advantage of this advance. Success in this work, implying the ability to learn modular, hierarchical solutions to problems, will greatly improve the scalability of Genetic Programming and its ability to solve complex problems. It will also have important ramifications for the field of artificial life.

Ph.D Research

Supervisor:

           Dr. Daryl Essam, School of ITEE, UNSW@ADFA, Canberra, Australia.

Co-supervisors:

          Ass. Prof. R.I (Bob) McKay,  Structure Complexity Lab, School of CS&E,                                                          Seoul National University, Seoul, Korea.

          Dr. Xuan Hoai Nguyen, Department of Vietnam Military Technical                                                              Academy, Hanoi, Vietnam.

 

 

Tuan-Hao Hoang’s Home Page