Information Technology Reference
In-Depth Information
Endocrine-Immune Network and Its Application
for Optimization
Hao Jiang, Tundong Liu , Jing Chen, and Jiping Tao
College of Information Science and Technology, Xiamen University,
Xiamen 361005, China
jianghao 1017@163.com, { ltd,taojiping } @xmu.edu.cn, chenjxmu@126.com
Abstract. A novel artificial immune network model (EINET) based on
the regulation of endocrine system is proposed. In this EINET for opti-
mization, several operators are employed or revised which aim at faster
convergence speed and better optimal solution. Further speaking, a new
operator, hormonal regulation, exerts a bidirectional regulatory mecha-
nism inspired from endocrine system, which undergoes elimination and
mutation according to hormone updating function, to increase the di-
versity of antibody population. And antibody learning is an evolution of
individuals through learning from memory antibody in immune network.
Then, a local search procedure called enzymatic reaction is utilized to
facilitate the exploitation of the search space and speed up the conver-
gence. To evaluate whether the proposed model can be directly extended
to an effective algorithm for solving combinatorial optimization prob-
lem, EINET-TSP algorithm is designed. Comparative experiments are
conducted using some benchmark instances from the TSPLIB, and the
results compared with the existing immune network applied to combi-
natorial optimization problem shows that the EINET-TSP algorithm is
capable of improving search performance significantly in solution quality.
Keywords: Artificial Immune Network, Hormonal regulation, Enzy-
matic reaction, Traveling Salesman Problem.
1 Introduction
Over the past few years, based on principles of the immune system, a new
paradigm, called artificial immune system (AIS), has been employed for develop-
ing interesting algorithms in many fields such as pattern recognition, computer
defense, optimization, and others. Artificial Immune Network (AIN) is inspired
from immune network theory originally proposed by Jerne [1], which has been
one of the most important immune theories. In recent years, s a large number of
AINs have been developed, two popular approaches are RLAIS model [2],which
is modified from a earlier version named AINE, and aiNet model presented by
de Castro [3,4,5]. The aiNet enhanced the clonal selection algorithm (CLON-
ALG) [6] by combining it with immune network theory. In a subsequent work,
 
Search WWH ::




Custom Search