TY - GEN
T1 - An immigrants scheme based on environmental information for genetic algorithms in changing environments
AU - YU, Xin
AU - TANG, Ke
AU - YAO, Xin
PY - 2008/6
Y1 - 2008/6
N2 - Addressing dynamic optimization problems (DOPs) has been a challenging task for the genetic algorithm (GA) community. One approach is to maintain the diversity of the population via introducing immigrants. This paper intensively examines several design decisions when employing immigrants schemes, and from these observations an environmental information-based immigrants scheme is derived for GAs to deal with DOPs. In the scheme, the environmental information (e.g., the allele distribution over the population in this paper) from previous generation is used to create immigrants to replace the worst individuals in the current population. In this way, the introduced immigrants are more adapted to the changing environment. A hybrid scheme combining immigrants based on current environmental information and its complementation is also proposed in this paper to address different degrees of changes. Experimental results validate the efficacy of the proposed environmental information-based and hybrid environmental information-based immigrants schemes. © 2008 IEEE.
AB - Addressing dynamic optimization problems (DOPs) has been a challenging task for the genetic algorithm (GA) community. One approach is to maintain the diversity of the population via introducing immigrants. This paper intensively examines several design decisions when employing immigrants schemes, and from these observations an environmental information-based immigrants scheme is derived for GAs to deal with DOPs. In the scheme, the environmental information (e.g., the allele distribution over the population in this paper) from previous generation is used to create immigrants to replace the worst individuals in the current population. In this way, the introduced immigrants are more adapted to the changing environment. A hybrid scheme combining immigrants based on current environmental information and its complementation is also proposed in this paper to address different degrees of changes. Experimental results validate the efficacy of the proposed environmental information-based and hybrid environmental information-based immigrants schemes. © 2008 IEEE.
UR - http://www.scopus.com/inward/record.url?scp=55749084033&partnerID=8YFLogxK
U2 - 10.1109/CEC.2008.4630940
DO - 10.1109/CEC.2008.4630940
M3 - Conference paper (refereed)
SN - 9781424418237
SP - 1141
EP - 1147
BT - 2008 IEEE Congress on Evolutionary Computation, CEC 2008
ER -