In the design and management of modern systems human decision makers have to deal with an ever more increasing complexity in terms of data volumes, network connectivity, and nonlinearity of models. At the same time decisions often have to be made fast. To deal with these challenges decision support software has become a potent and necessary tool for design and decision making. It makes it possible to explore large spaces of alternatives in relative short time and to perform complex, simulation-based, predictions of performance measures. Classical algorithms for decision support, however, reach their limits for problems with extremely large numbers of alternatives and/or criteria. Another difficulty arises from the nonlinearity and size of models used in decision analysis and optimization and the difficulty to precisely capture user preferences in automated search. The development of algorithmic methods that can deal with more complex settings of optimization and decision support is thus an ongoing challenge.

The methodological focus of the MODA research group is on natural computing and computational intelligence (evolutionary algorithms, swarm intelligence, machine learning), which are versatile computational methods, often inspired by natural adaptation processes, and have favorable properties such as fault tolerance, self-adaptation, set-orientation, parallelizability, and scalability. Most computational intelligence methods are based on some collective search or learning paradigm. They can thus readily exploit the available computational resources by means of parallel problem solving techniques. As most computational intelligence methods are heuristics and cannot guarantee optimal solutions it is attractive to design combinations of natural computing methods with exact methods so called hybrid methods that can in many cases yield local convergence guarantees while preserving the advanced global search performance of natural computing methods. In terms of the problem domain the key focus of the research in the MODA group is to develop and disseminate theoretical results and algorithms for solving optimization and decision analysis problems with multiple objectives and constraints. Our current publicly funded research projects are:

  • EXCELLENT BUILDINGS: Excellent Buildings via Forefront Multidisciplinary Optimization: Lowest Energy Use, Optimal Spatial and Structural Performance
  • DELIVER: Distributed Coordination for Continuous Planning (EUREKA)
  • RODEO: Robust Design Optimization (NWO)
  • BETNET: The evolution of stochastic heterogeneous networks as bet-hedging adaptations to fluctuating environments (NWO)
  • PHARMA-IT: Multiobjective Optimization in De-Novo Drug Discovery.
  • NiCaiA: Combining artificial intelligence and Multiobjective Optimization
  • WaterNet: Multicriteria design of robust and energy efficient water distribution networks.
  • SIMO: Theoretical foundations of set-indicator based multi-objective optimization.
  • EMONA: Evolutionary Multicriteria Optimization for Network Analytics and Prediction