Information Technology Reference
In-Depth Information
n-TSP
IA
Definition of problem
parameters such as the
cities and salesmen
Problem presentation
Define the information
of salesmen and cities
Making initial salesman
The salesmen are made at
random or using a memory
of previous results
Initialization of each
immune cell set
Deletion and addition of
immune cell sets on
each salesman agent
Making behaviors
Candidate behaviors
are made at random
Evaluation of a behavior
If the salesman has the applicable
behavior, then it can act. If the target
city has already been visited by other
salesman, then each action is
defined by the MHC
Selection of next cities
by salesman agents
Proposed immune algorithm (IA) for solving the n-travelling salesman problem (TSP). Each
immune cell set is composed of three kinds of cells, called a macrophage, a B, and a T cell.
Figure 7.13
Corresponding components of immune algorithm and n-TSP.
7.9 Other Applications
7.9.1
Developing Associative Memories
Researchers (Smith et al., 1996) argued that the immunological memory is a mem-
ber of the family of sparsely distributed memories, and it derives associative and
robust properties from a sparse and distributed nature of sampling.
Figure 7.14 illustrates the formation of immune memory (as the concentration
level of various immune cells) during the primary and secondary responses.
AINs have been applied to create an associative memory model (Singh and
h ayer, 2001). Associative memory is used to remember patterns and enable fast
and eff ective recall of those patterns. h e authors implemented two mechanisms
defi ned by Abbattista et al. (1996), namely, the immune system metadynamics and
the immune recruitment mechanism. A population of points in the space is defi ned,
which compete to recruit items from the training population. h ese result in cluster-
ing areas on the surface space that, in eff ect, store the patterns being learned.
7.9.2
Applications in Games
Many AIS researchers (Varela et al., 1988) talked about the inherent capability of the
immune networks for machine learning. Perelson (1989) and Cooke and Hunt (1995)
 
Search WWH ::




Custom Search