|Topic:||Evoluční algoritmy s nepřímou reprezentací|
|Supervisor:||Ing. Jiří Kubalík, Ph.D.|
|Announce as:||Bakalářská práce, Semestrální projekt|
|Description:||An evolutionary algorithm with an indirect representation is an efficient approach to solving hard constrained optimization problems. The “indirect representation” means that the evolved individuals do not encode the solution parameters directly. Instead, each solution is represented by a so-called priority list that encodes the order in which solution components are one by one added to the final solution using constructive heuristics designed for the problem at hand. This way, a portion of the optimization task is delegated to the heuristics. This approach is particularly suitable for solving combinatorial optimization problems.
This work aims to extend this approach with a mechanism that identifies high-level solution components on the fly during the evolutionary process and uses them as new elements that can be used in the priority lists. In particular, the challenges are how to identify possibly useful high-level components, how to deal with variable-length priority lists, how to design genetic operators, etc.
|Bibliography:||Jiří Kubalík, Petr Kadera, Václav Jirkovský, Lukáš Kurilla, Šimon Prokop:
Plant Layout Optimization Using Evolutionary Algorithms. HoloMAS 2019: 173-188
Júnior Bonfim et al.: A biased random-key genetic algorithm using dotted board model for solving two-dimensional irregular strip packing problems. 2020 IEEE Congress on Evolutionary Computation (CEC), DOI: 10.1109/CEC48606.2020.9185794, 2020.