Many of the existing network theory based artificial immune systems have been applied to data clustering. The formation of artificial lymphocyte (ALC) networks represents potential clusters in the data. Although these models do not require any user specified parameter of the number of required clusters to cluster the data, these models do have a drawback in the techniques used to determine the number of ALC networks. This paper discusses the drawbacks of these techniques and proposes two alternative techniques which can be used with the local network neighbourhood artificial immune system. The end result is an enhanced model that can dynamically determine the number of clusters in a data set.

Abstract:The natural immune system (NIS) protects the body against unwanted foreign material (non-self cells) that could damage the body (self cells). The NIS can be modeled into an artificial immune system (AIS) to detect any non-self patterns in a non-biological environment. Detectors in the NIS can change from their initial mature status to memory status detectors or to annihilated status. A memory detector is a detector that frequently detects non-self cells and is a general detector for a subset of non-self cells. The NIS uses these memory detectors in a faster response to non-self cells. The purpose of this paper is to present the genetic artificial immune system (GAIS) which evolves these non-self detectors and determine their state using a life counter function. Only detectors with mature or memory status are used to detect non-self. Thus, the number of detectors is dynamically determined by the life counter function. In the paper GAIS is applied to different classification problems.

No abstract

The network theory in immunology inspired the modeling of network based artificial immune system (AIS) models for data clustering. Current network based AIS models determine the network connectivity between artificial lymphocytes (ALCs) by measuring the spatial distance between these ALCs against a distance threshold or by grouping ALCs into sub-networks. This paper discusses alternative network topologies to determine the network connectivity between ALCs and the advantages of using these network topologies. The local network neighborhood AIS model is then proposed as a network based AIS model which uses an index-based ALC neighborhood to determine the network connectivity between ALCs. The proposed model is compared to existing network based AIS models which are applied to data clustering problems. Furthermore, a sensitivity analysis is also done on the proposed model to investigate the influence of the model's parameters on the quality of the clusters. The paper also gives a formal definition of data clustering and discusses the performance measures used to determine the quality of clusters.

No abstract