Thesis structure
1.3. Thesis structure¶
Chapter 2 introduces selected topics related to Learning Classifier Systems and handling of the real-valued input. The evolution of a system capable of predicting the consequences of executed actions is emphasized. Moreover, the experimentation process is described, including the statistical verification of obtained results and the description of key performance indicators alongside benchmarking problems. Chapter 3 proposes two procedures for dealing with a real-valued input. The first one involves a novel system, representing the environmental state by the interval predicates, while the other transforms input signal before an actual algorithm perceives it. Both Chapter 4 and Chapter 5 focus on potential issues spotted when processing real-valued data - optimizations for a more efficient knowledge collection process and the problems when obtaining the reward feedback requires multiple steps. Finally, Chapter 6 concludes the thesis and presents potential future research directions.