Project topics and materials
Get Complete Project Material File(s) Now! »
Idiotypic NetworkModels
A number of different theoretical network based models have been proposed by immunologists to formulate and capture the characteristics and interactions of the natural immune network system.
One of these theoretical network based models was proposed by Farmer et al. [49]. Farmer et al. exploited the fundamental concepts of the network theory as proposed by Jerne [97] and proposed a simple model to simulate the dynamics of the natural immune network system and its memory capability [49]. Perelson also proposed a model to simulate the dynamics and production of a network based immune network system [148]. The theoretical model proposed by Farmer et al. is discussed next, since the earliest work in artificial immune systems are based on this model. The rest of the section discusses different network theory inspired AIS models.
The theory of clonal selection as part of the process of affinity maturation assumes that all immune responses are activated by encountered antigens. As explained in section 3.5, antigens select those lymphocytes with which the antigens have the highest binding affinity, resulting in clonal proliferation and somatic hyper mutation of the selected lymphocytes. As a result of somatic hyper mutation on the clones, the variable regions of the clones can become antigenic and invoke an immune response from neighbouring lymphocytes (as discussed in section 3.6). The recognition of idiotopes results in interconnected neighbouring lymphocytes, forming an idiotypic network.
Thus, lymphocytes in a network co-stimulate and/or co-suppress each other in reaction to an antigen. Therefore a lymphocyte is not only stimulated by an antigen, but also by neighbouring lymphocytes (as discussed in section 3.6). This results in the annihilation of some lymphocytes and the introduction of mutated lymphocyte clones into the population of lymphocytes. Highly stimulated lymphocytes remain part of the population whereas less stimulated lymphocytes are replaced/removed from the population. The population of lymphocytes is dynamic in such a way that the concentration of antibodies/lymphocytes at different points in time differ. In order to formulate the change in concentration of the population, based on the stimulation of the lymphocytes, Farmer et al. [49] identified three factors which influence the stimulation of a lymphocyte.
Idiotypic Network Topologies
The formation of idiotypic networks between lymphocytes (or their corresponding antibodies) can be defined by different network topologies. In the preceding section on network based artificial immune systems, the network interaction or network formation between artificial lym-similar artificial lymphocytes in sub-networks. The former is normalised with a network affinity threshold to determine the network links between the artificial lymphocytes and the latter utilises a clustering algorithm. There are, however, alternative and less familiar network topologies to determine the possible interactions in an idiotypic network of lymphocytes. The different theoretical approaches to determine the possible interactions in an idiotypic network are discussed next. Each of these network topologies specifies the interconnections between lymphocytes and the binding strength of these connections.
The Linear Topology: Lymphocytes in the linear topology are positioned as a sequence of different idiotypic levels of interaction. The linear topology was introduced by Richter and proposed as a chain-reaction between lymphocytes at different idiotypic levels [152, 153]. Figure 4.4 illustrates the linear topology of an idiotypic network with l idiotypic levels.
1 Introduction
1.1 Motivation
1.2 Objectives
1.3 Methodology
1.4 Contributions
1.5 Thesis Outline
2 Clustering and Quality Measures
2.1 Data Clustering
2.2 Similarity Measures
2.3 Clustering Algorithms
2.4 Cluster Quality Validation
2.5 Cluster Quality in Dynamic and Uncertain Environments
2.6 Outlier Detection and Analysis
2.7 Alternative Computational Models for Clustering
2.8 Conclusion
3 The Natural Immune System
3.1 Classical View
3.2 Antibodies and Antigens
3.3 The White Cells
3.4 Immunity Types
3.5 The Process of Affinity Maturation
3.6 The Network Theory
3.7 The Danger Theory
3.8 The Dendritic Cell System
3.9 Conclusion
4 Artificial Immune Systems
4.1 A Basic AIS Framework
4.2 Representation of Antigens and Antibodies
4.3 Affinity as Quality Measure
4.4 Negative Selection Models
4.5 Clonal Selection Models
4.6 Idiotypic Network Models
4.7 Idiotypic Network Topologies
4.8 Danger Theory Models
4.9 Conclusion
5 A Local Network Neighbourhood Artificial Immune System with Application to Unsupervised Data Clustering
5.1 The Algorithm
5.2 Initialising an ALC and the ALC population
5.3 Reacting to an Antigen
5.4 Adapting the ALCs in a Local Network Neighbourhood
5.5 Similarities and Differences with Other Network based AIS Models
5.6 Time Complexity of LNNAIS
5.7 Experimental Results and Analysis
5.8 Conclusion
6 Dynamically Determining the Number of Clusters Found by a Local Network Neighbourhood Artificial Immune System
6.1 Dynamic Data Clustering Methods
6.2 Dynamic Clustering Techniques for LNNAIS
6.3 Time Complexity of SDOT and IPT
6.4 Experimental Results
6.5 Influence of LNNSDOT Parameters
6.6 Conclusion
7 Data Clustering in Non-stationary Environments using a Local Network Neighbourhood Artificial Immune System
7.1 Clustering Performance Measures for Non-stationary Environments
7.2 Data Migration Types
7.3 Generating Artificial Non-stationary Data
7.4 Sensitivity of LNNAIS parameters
7.5 Experimental Results
7.6 Conclusion
8 Conclusion
8.1 Summary
8.2 Future Research
References
