since 2001-10-31
Gene contains not only the squence to control the phenotype which account for 4% of whole DNA squence, but also the squence to control the behavior of the ogranism. This is the program to control the life.
Corresponding to this hyptothesis: if the coding in GA is not only translated into phenotype, but also into behavior feature(or say, instinct), in individual in GA will be an agent. Such a behavior feature can be depicted by "if-then". We can name it as "Behavior Evolutional Genetic Algorithm"(BEGA)
( Jie Bao 2000-06-12)
The kernel of BEGA is not what structure of the chromosome, but a hierarchical behavior formation by a linear encoding.
Two major kinds of hierarchy in GA: in encoding stage (which is BEGA) and in Selection stage (which is sociobiological group)
(Jie Bao 2003-01-15)
Actually the research is a combination of MAS and GA. the population is a MAS with all agent want to maximize their expected utility. However, their utility is not expressed directly via a table, but be translated from their decisions to real number(fitness). They are searching the optimal decision using a team search and share common information (cross).
Some Approaches
(Jie Bao 2003-03-13)
2001-10-31: Description for BEGA plan
2001-11-22:
Behavior Evolutional Genetic Algorithm
2001-12-13: Information Filtering based on Behavior Evolutional Genetic Algorithm(paper) Code Data
2002-02-04: GP software: GPC++
Term Project Plan for CS 572
--
Behavior Evolutional Genetic Algorithm
moved to here
[Return to Jie Bao's Homepage]