3rd year Java and Genetic Programming Projects

Java based Genetic Programming Demonstration

Java based distributed Genetic Programming Demonstration

Parallel Distributed Genetic Programming Implementation

BACKGROUND:
Evolutionary Computation (EC) is a field the main objective of which is to try and imitate reproduction genetics and natural selection to solve engineering problems such as optimisation, search and machine learning. Genetic Programming (GP) is an EC technique which allows the automatic evolution of programs that solve a given problem.


Java based Genetic Programming Demonstration

SUPERVISOR: W. B. Langdon

COURSE SUITABILITY: ?

BRIEF DESCRIPTION:
The project involves the demonstration of GP running over the internet. The system must support execution on a number of different computer types (eg PC, Sun, Alpha) running as Java applets within an internet browser such as HotJava. At least one demonstration GP problem (eg the Artificial Ant following the Santa Fe trail) must be solved. The system must be user friendly and especially be a "good guest" when run on remote internet hosts.

The project must include graphical interfaces to display GP progress, execution of intermediate and final evolved programs. Possible extensions include GUIs to create or extend GP problems, including specifying function/terminal sets, termination criterion, fitness functions. Other features, such as saving GP populations, recovering from link or machine failures, may require code to be run as a Java application rather than applet. Unified source code supporting both application and applet are required.

EQUIPMENT: Unix workstation, PC

LANGUAGES AND/OR OTHER SOFTWARE:
Java software in the public domain will be required
HTML, possibly cgi scripts

COMMENTS:
Early in the project the student will be expected to define which (if any) public domain GP/GA java code is to be used and the GUI interfaces.


Java based distributed Genetic Programming demonstration

SUPERVISOR: W. B. Langdon

COURSE SUITABILITY: ?

BRIEF DESCRIPTION:
The project involves the demonstration of distributed GP running over the internet. The system must support parellel execution on hetrogenous computers (eg PC, Sun, Alpha) running as Java applets under HotJava, Netscape or InternetExplorer. At least one (eg the Artificial Ant following the Santa Fe trail) demonstration GP problem must be solved. The system must be user friendly and especially be a "good guest" when run on remote internet hosts.

The project could include multiple models of distribution, (population, CPU load, etc), dynamic addition and removal of internet computers, graphical interfaces to display GP progress, execution of intermediate and final evolved programs, GUIs to create or extend GP problems, including specifying function/terminal sets, termination criterion, fitness functions. Other features, such as saving GP populations, recovering from link or machine failures, may require code to be run as a Java application rather than applet. Unified source code supporting both application and applet are required.

EQUIPMENT: Unix workstation, PC

LANGUAGES AND/OR OTHER SOFTWARE:
Java software in the public domain will be required
HTML, possibly cgi scripts

COMMENTS:
Early in the project the student will be expected to define the model(s) of distribution to be supported, which (if any) public domain GP/GA java code is to be used and the GUI interfaces.


Parallel Distributed Genetic Programming Implementation

SUPERVISOR: W. B. Langdon

COURSE SUITABILITY: ?

BRIEF DESCRIPTION:
Parallel Distributed Genetic Programming (PDGP) is a version of genetic programming in which the usual tree programs are replaced by a network fixing the structure of the program and allowing code re-use by multiple links.

The project involves the implementation and demonstration of PDGP in Java. At least one (eg the Artificial Ant following the Santa Fe trail) demonstration GP problem must be solved.

The project could include graphical interfaces to display PDGP progress, execution of intermediate and final evolved programs, GUIs to create or extend PDGP problems, including specifying network grid, function/terminal sets, termination criterion, fitness functions. Other features, such as saving populations, may require code to be run as a Java application rather than applet. Unified source code supporting both application and applet are required.

EQUIPMENT: Unix workstation or PC

LANGUAGES AND/OR OTHER SOFTWARE:
Java software in the public domain may be required
HTML, possibly cgi scripts

COMMENTS:
Early in the project the student will be expected to define if any public domain GP/GA java code is to be used and the GUI interfaces.