Group for computational intelligence and mathematical optimization
The group conducts research in the field of artificial intelligence, with a focus on the theory, applications, and development of computational intelligence and mathematical optimization. Computational intelligence (CI), as one of the areas of artificial intelligence, initially encompassed neural networks, fuzzy systems, and evolutionary algorithms, while today CI encompasses an increasingly broad range of methods. Mathematical optimization concerns the study of problems in which the objective is to maximize or minimize a real-valued function over real or integer variables belonging to a specified domain, allowing a large number of practical problems to be formulated in this way. The group's research activities cover a broad range of topics, including mathematical optimization, machine learning, and fuzzy logic. The research is based on computational intelligence techniques and methods for modeling and solving problems arising in complex systems.
The group has achieved significant results in theoretical research, the development of models and algorithms for mathematical optimization problems, and their practical applications. Exact solutions to problems are primarily obtained by implementing mathematical models in some of the most widely used modern general-purpose solvers, such as CPLEX, GUROBI, and popular SAT solvers. For certain problems, specialized exact solvers have also been developed using the branch-and-bound method or model partitioning techniques (by creating supermodels and submodels). The group has also been highly successful in developing metaheuristic methods for solving highly complex problems. In addition to population-based metaheuristics (e.g. evolutionary algorithms, particle swarm optimization, etc.), the group's researchers successfully apply metaheuristics based on local search methods (e.g. Variable Neighborhood Search (VNS), Greedy Randomized Adaptive Search Procedure (GRASP), tabu search, etc.). The group also bases its research on combining several of the methods described above, both exact and metaheuristic, thereby obtaining hybrid methods that have often proved to be the most effective in practice. Parallelization is also employed in the group's research to improve performance.
Some information about the work of this group is available at: