In the resource limited artificial immune system (RLAIS), because the network granularity is determined by the network affinity threshold (NAT) and the initialization value of NAT is obtained by calculating the distance between the antigens each other, the NAT doesn't reflect the network evolution process. The computation of the stimulation level at closer distance doesn't sufficiently reflect its advantage and is too sensitive to the change in the distance. The adoption of pure clone selection and random change operation strategy can impair the convergence of the public and the stability. By analyzing the disadvantages of RLAIS, the modified resource limited artificial immune system (MRLAIS) was proposed. In MRLAIS, during the network evolution process the adaptation threshold value is computed again in each iterative to better characterize the state of affinity of the antibody at that time. A stimulation function is selected that which sufficiently incarnate the advantage of the stimulation level when the antibodies distance is small and is not too sensitive to smaller distances. Finally, a resource allocation function is selected to make the network allocate the antibody more reasonably. The experimental stimulation shows that the MRAIS has the faster evolution speed and the better structure stability.
[1]
Claudia Eckert,et al.
On the Use of Hyperspheres in Artificial Immune Systems as Antibody Recognition Regions
,
2006,
ICARIS.
[2]
Lois Boggess,et al.
ARTIFICIAL IMMUNE SYSTEM CLASSIFICATION OF MULTIPLE- CLASS PROBLEMS
,
2002
.
[3]
Fernando José Von Zuben,et al.
omni-aiNet: An Immune-Inspired Approach for Omni Optimization
,
2006,
ICARIS.
[4]
Fabio A. González,et al.
A Scalable Artificial Immune System Model for Dynamic Unsupervised Learning
,
2003,
GECCO.
[5]
Jonathan Timmis,et al.
A resource limited artificial immune system for data analysis
,
2001,
Knowl. Based Syst..
[6]
Jon Timmis,et al.
Computer Science at Kent Assessing the performance of the resource limited artificial immune system AINE
,
2001
.