ALGO talks with Andreia Guerreiro
On 2 July, 18h00, Andreia Guerreiro, Researcher at INESC-ID, Lisbon, will give a talk entitled " Exploiting an Integer Linear Programming Formulation for the Hypervolume Subset Selection Problem".
Set-quality indicators, which map a set of points into a scalar value are a convenient way of assessing the quality of a set of solutions in multiobjective optimization. The hypervolume indicator is one of the most commonly used quality indicators. Due to its theoretical properties, it is frequently used within the (environmental) selection step in Evolutionary Multiobjective Optimization Algorithms (EMOAs). In such a case, selection may be viewed as a Hypervolume Subset Selection Problem (HSSP) that consists of selecting a subset of k points out of a set of n points that maximizes the hypervolume indicator. However, already with 3 objectives, HSSP-based selection in EMOAs is usually solved either for particular cases such as k=n-1, or with greedy algorithms. Although a few exact algorithms exist to compute the HSSP for k<n-1, faster algorithms are required to make its integration in EMOAs practical. In this talk, a new integer linear programming formulation for the HSSP is presented, as well as an algorithm that exploits such formulation to considerably speed up the computation of the HSSP. Such algorithm makes HSSP-based selection in EMOAs more practical, at least for 3 objectives.
Andreia P. Guerreiro is currently a junior researcher and a member of the Automated Reasoning and Software Reliability group at INESC-ID in Lisbon. She obtained her Masters degree in Information Systems and Computer Engineering from Instituto Superior Técnico, and her PhD degree in Information Science and Technology from the University of Coimbra. Her research work is concerned with approximation algorithms for multiobjective combinatorial optimization, the development and analysis of efficient indicator-based selection algorithms for evolutionary multiobjective optimization, as well as algorithms for performance assessment in multiobjective optimization.