TY - GEN
T1 - Elitistic evolution
T2 - 9th International Conference on Adaptive and Natural Computing Algorithms, ICANNGA 2009
AU - Jiménez, Francisco Viveros
AU - Mezura-Montes, Efrén
AU - Gelbukh, Alexander
PY - 2009
Y1 - 2009
N2 - A new evolutionary algorithm, Elitistic Evolution (termed EEv), is proposed in this paper. EEv is an evolutionary method for numerical optimization with adaptive behavior. EEv uses small populations (smaller than 10 individuals). It have an adaptive parameter to adjust the balance between global exploration and local exploitation. Elitism have great influence in EEv' proccess and that influence is also controlled by the adaptive parameter. EEv' crossover operator allows a recently generated offspring individual to be parent of other offspring individuals of its generation. It requires the configuration of two user parameters (many state-of-the-art approaches uses at least three). EEv is tested solving a set of 16 benchmark functions and then compared with Differential Evolution and also with some well-known Memetic Algorithms to show its efficiency. Finally, EEv is tested solving a set of 10 benchmark functions with very high dimensionality (50, 100 and 200 dimensions) to show its robustness.
AB - A new evolutionary algorithm, Elitistic Evolution (termed EEv), is proposed in this paper. EEv is an evolutionary method for numerical optimization with adaptive behavior. EEv uses small populations (smaller than 10 individuals). It have an adaptive parameter to adjust the balance between global exploration and local exploitation. Elitism have great influence in EEv' proccess and that influence is also controlled by the adaptive parameter. EEv' crossover operator allows a recently generated offspring individual to be parent of other offspring individuals of its generation. It requires the configuration of two user parameters (many state-of-the-art approaches uses at least three). EEv is tested solving a set of 16 benchmark functions and then compared with Differential Evolution and also with some well-known Memetic Algorithms to show its efficiency. Finally, EEv is tested solving a set of 10 benchmark functions with very high dimensionality (50, 100 and 200 dimensions) to show its robustness.
UR - http://www.scopus.com/inward/record.url?scp=78650746496&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-04921-7_18
DO - 10.1007/978-3-642-04921-7_18
M3 - Contribución a la conferencia
SN - 3642049206
SN - 9783642049200
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 171
EP - 182
BT - Adaptive and Natural Computing Algorithms - 9th International Conference, ICANNGA 2009, Revised Selected Papers
Y2 - 23 April 2009 through 25 April 2009
ER -