- Supervisions
- Publications
- Tutorials
- Workshops
- Tech Reports
- Locust Swarms
- Commonality-Based Local Search
- Educational Tools
- Financial Modelling
- Knowledge Translation and Information Organization
Locust Swarms
Locust Swarms are a multi-optima particle swarm that involve two phases: Scouts (for coarse search) and Locusts (for greedy search). Early results include best-known results on the "FastFractal" problem used in the CEC2008 Large Scale Global Optimization Competition. Current research is focusing on the role of "dimension reductions" on non-decomposable problems -- e.g. BBOB. [read more...]
Commonality-Based Local Search
Local search techniques (e.g. hill climbing, simulated annealing, tabu search, genetic algorithms, evolutionary strategies, etc) all employ change operators. A change operator creates a new candidate solution from an existing candidate solution(s) by keeping some parts and changing some parts. Most change operators focus on what is changed -- e.g. a two-opt swap targets two edges to be changed. However, what is kept can also have an important effect on search performance.
The commonality hypothesis suggests that the common components of above-average solutions are above average -- these parts of the solution should be kept. By restricting changes to the uncommon components, the likelihood of an improving change can be increased. Embedded within the change operator, the benefits of preserving common components are independent of the control and selection strategies of the local search technique. These benefits have been isolated and demonstrated in genetic algorithms and simulated annealing. [read more...]
Educational Tools
I am a member of CLOE -- The Co-operative Learning Object Exchange. My primary research interest in this area is to develop new educational tools for computer science education. I am also a member of ACM SIGCSE -- the ACM Special Interest Group on Computer Science Education. [read more...]
Financial Modelling
Traditional economic models are better suited to measuring and/or explaining outcomes rather than modelling dynamic trajectories. These dynamic trajectories are important for the development and testing of intervention strategies. My research in this area explores the use of multi-agent systems to model and simulate the dynamics of financial markets. [read more...]
Knowledge Translation and Information Organization
As part of my research interest in educational tools, I am also interested in the effective and efficient organization and presentation of information through the internet. I am particularly interested in designs for accessibility. [read more...]
