Algorithm Design for Multi-Objective Problems

Multi-objective optimisation refers to an optimisation scenario having more than one objective function to be considered. Evolutionary algorithms have been found well-suited to multiobjective optimisation due to their population-based property of achieving an approximation of the Pareto optimal front in a single run. My research interests in this regard include developing a co-evolutionary search framework that makes multiple selection criteria complement each other and adapting the weights in MOEA/D to make it well-suited for any Pareto front shapes.

• Selected work

• [ECJ20] An adaptive weight update approach that enables the MOEA/D method to work well for any Pareto front shape. [Read More]

• [TEVC16] An algorithm framework which co-evolves two populations based on Pareto criterion and non-Pareto criterion respectively to complement each other. [Read More]

• [TEVC14] Developing a stable matching between subproblems and solutions in the decomposition-based approach.

• [ECJ14] Utilising the minimum spanning tree to maintain diversity of the population during the evolutionary process.