TY - CONF
T1 - Optimization of Fuzzy Systems Using Group-Based Evolutionary Algorithm
AU - Chang, Jyh-Yeong
AU - Han, Ming Feng
AU - Lin, Chin-Teng
PY - 2012
Y1 - 2012
N2 - This paper proposes a group-based evolutionary algorithm (GEA) for the fuzzy system (FS) optimization. Initially, we adopt an entropy measure method to determine the number of rules. Fuzzy rules are automatically generated from training data by entropy measure. Subsequently, the GEA is performed to optimize all the free parameters for the FS design. In the evolution process, a FS is coded as an individual. All individuals based on their performance are partitioned into a superior group and an inferior group. The superior group, which is composed of individuals with better performance, uses a global evolution operation to search potential individuals. In the inferior group, individuals with a worse performance employ the local evolution operation to search better individuals near the current best individual. Finally, the proposed FS with GEA model (FS-GEA) is applied to time series forecasting problem. Results show that the proposed FS-GEA model obtains better performance than other algorithm.
AB - This paper proposes a group-based evolutionary algorithm (GEA) for the fuzzy system (FS) optimization. Initially, we adopt an entropy measure method to determine the number of rules. Fuzzy rules are automatically generated from training data by entropy measure. Subsequently, the GEA is performed to optimize all the free parameters for the FS design. In the evolution process, a FS is coded as an individual. All individuals based on their performance are partitioned into a superior group and an inferior group. The superior group, which is composed of individuals with better performance, uses a global evolution operation to search potential individuals. In the inferior group, individuals with a worse performance employ the local evolution operation to search better individuals near the current best individual. Finally, the proposed FS with GEA model (FS-GEA) is applied to time series forecasting problem. Results show that the proposed FS-GEA model obtains better performance than other algorithm.
KW - PARTICLE-SWARM OPTIMIZATION; DIFFERENTIAL EVOLUTION; CONTROLLER-DESIGN; GENETIC ALGORITHM; INFERENCE SYSTEMS; NETWORK
U2 - 10.1007/978-3-642-34487-9_36
DO - 10.1007/978-3-642-34487-9_36
M3 - Paper
SP - 291
EP - 298
ER -