This thesis describes new evolutionary artificial intelligence methods suitable for solving complex tasks. These include planning, optimization, decision, prediction and other problems. All of these are tasks which an intelligent human being can quickly learn to solve, yet they cannot be solved by machines in reasonable time. For this type of problems usually no analytical method or algorithm exists. These challenges represent the domain for artificial intelligence. This work concentrates on evolutionary methods of artificial intelligence based on genetic algorithms. Specifically grammatical evolution and differnetial evolution are described. The first part of this thesis describes the principles of genetic algorithms especially those used in grammatical evolution. Later the grammatical evolution method is described. Grammatical evolution is a genetic algorithm extended with a context-free grammar processor. This enables it to generate structured strings in an arbitrary language defined by a regular or context-free grammar. Second part of this work focuses on description of a generic computational system, which enables user-friendly control of grammatical evolution. The architecture of the system is thoroughly described. It composes of a computation service, database server and completely separated user interface. Also the problems solved using this system are described. These include symbolic regression, classification and generation of combinatorial logic circuits. All of these tasks were solved using the described implementation.
| Identifer | oai:union.ndltd.org:nusl.cz/oai:invenio.nusl.cz:249236 |
| Date | January 2009 |
| Creators | Popelka, Ondřej |
| Source Sets | Czech ETDs |
| Language | Czech |
| Detected Language | English |
| Type | info:eu-repo/semantics/doctoralThesis |
| Rights | info:eu-repo/semantics/restrictedAccess |
Page generated in 0.0014 seconds