Combinatorial Optimization and Decision Support (Belgium, Coordinator)
Description: The Combinatorial Optimisation and Decision Support (CODeS) within the Department of Computer Science investigates the construction, the behaviour and the application of metaheuristics for combinatorial optimisation, centralised and distributed, at the common frontier of artificial intelligence and optimisation. Running and completed projects show a diversity of application domains. Planning, scheduling and timetabling search in large datasets and optimisation in agent based systems are the expertise of the group. Application domains cover real world scheduling and timetabling problems, e-learning and bio-informatics. We pursue generic approaches towards models as well as algorithms. The group counts 20+ people among which three staff members and two postdoctoral collaborators.
The group consists of two subgroups. The subgroup in Kortrijk was established in the fall of 2005. Its subject is the study of heuristic combinatorial optimisation and automated learning and its applications in distributed systems, scheduling, timetabling, and e-learning and bioinformatics. The members of this sub team often work in a strongly interdisciplinary setting, e.g. with physicists, biologists, chemists, language, psychologists and educational scientists. The group at KaHo Sint-Lieven started in 1997. It focuses on combinatorial optimisation problems that companies, often SMEs, identify as hard to solve. Applications include healthcare personnel rostering, production scheduling, timetabling, vehicle routing. The developed approaches vary from exact mathematical methods to modern heuristics and learning approaches.
The head of the group, Patrick De Causmaecker, holds a PhD in elementary particle physics from the Katholieke Universiteit Leuven (1983). In his PhD he established a calculus that revolutionised the evaluation of cross sections in high energy and elementary particle physics. After more than twenty years, his papers on this subject are still regularly cited. After he switched to the then booming field of information processing in 1984, he has been educating students at various levels in computer science. From 1994 on he has successfully conducted research in scheduling and optimisation, especially when human time spending was involved. This research was in close cooperation with a multitude of small and medium sized companies specialised in planning and scheduling for production, transport, education and medical care. Recently, he embarked on optimised scheduling in distributed systems and networks as a way to improve the quality of experience. At present he is especially interested in structures and methods to support the development of software systems for complex computational applications and the use of heuristics therein. Special attention goes to developments in meta heuristics and hyper heuristics. In 2002-2003 he spent one year with the ASAP (Automated Scheduling, optimisAtion and Planning) group at the University of Nottingham, UK. He is reviewer for major journals and conferences, and is member of the steering committee of the PATAT series of conferences. From October 2005 onwards he is a professor of informatics at Katholieke Universiteit Leuven, Campus Kortrijk.
Role of the team in the project:
The CODeS research group brings in strong expertise in statistical analysis, machine learning, combinatorial optimization and agent based approaches for high performance computing. Through its close link with computer science, it will warrant the quality of the developed software modules from a software engineering point of view. Its experience with developing and testing optimization algorithms will allow delivering high quality robust systems resulting in near optimal prediction modules and strategies.