The Role of Hypermutation and Affinity Maturation in AIS Approaches to Clustering

The Role of Hypermutation and Affinity Maturation in AIS Approaches to Clustering

Waseem Ahmad (International College of Auckland, New Zealand) and Ajit Narayanan (Auckland University of Technology (AUT), New Zealand)
DOI: 10.4018/978-1-4666-8513-0.ch007
OnDemand PDF Download:
No Current Special Offers


In recent years, several artificial immune system (AIS) approaches have been proposed for unsupervised learning. Generally, in these approaches antibodies (or B-cells) are considered as clusters and antigens are data samples or instances. Moreover, antigens are trapped through free-floating antibodies or immunoglobulins. In all these approaches, hypermutation plays an important role. Hypermutation is responsible for producing mutated copies of stimulated antibodies/B-cells to capture similar antigens with higher affinity (similarity) measure and responsible to create diverse pool of solutions. Humoral-Mediated Artificial Immune System (HAIS) is an example of such algorithms. However, there is currently little understanding about the effectiveness of hypermutation operator in AIS approaches. In this chapter, we investigate the role of the hypermutation operator as well as affinity threshold (AT) parameters in order to achieve efficient clustering solutions. We propose a three-step methodology to examine the importance of hypermutation and the AT parameters in AIS approaches to clustering using basic concepts of HAIS algorithm. Here, the role of hypermutation in under-fitting and over-fitting the data will be discussed in the context of measure of entropy.
Chapter Preview


Clustering is one of the most intensively researched areas in the unsupervised learning and data mining disciplines. Clustering seeks to group similar data into clusters (groups) so that data instances within a group have maximum similarity while instances across different clusters have a high degree of dissimilarity. Clustering also depends on the nature of the data and the desired results or intuition (Tan, Steinbach, & Kumar, 2006). Therefore many clustering algorithms exist which use different induction principles. Recently, researchers have turned to natural phenomena for inspiration to develop new clustering algorithms. Underpinning this interest is an inclination with the nature and the emergence of complex learning behaviors and intelligence out of unstructured, unsupervised and decentralized processes, such as those in natural immune systems (NISs).

There is a rapidly growing interest in immune system inspired approaches to machine learning. Of particular interest is the way the human body responds to diseases and new pathogens as well as adapting to remain immune for long periods after a disease has been combated. Immune system processes consist of two phases: recognition of invaders, and response. It has been established that the NIS can adequately distinguish between threat and non-threat at a basic level. Also of interest is the way that the NIS can identify a self-cell (which is not to be reacted to) but which has been subsequently damaged in some way and might present a threat to the body (and which must be reacted to). In other words, the NIS is dynamic in that it can re-structure (re-classify or re-cluster) in the light of new information so that it provides protection against not only outside invaders but also inside dangers. All these NIS concepts, if carefully and systematically used, could confer great benefit in the area of machine learning. Apart from above mentioned characteristics, NISs also demonstrate the following capabilities:

  • Learning: An NIS continuously learns and adapts from the pathogens to trigger an appropriate immune response.

  • Diversity: An NIS consists of various cells and organs, which help to mount an immune response when seen or unseen pathogens are encountered.

  • Specialization/Generalization: An NIS has capabilities of specialization and generalization through the presence of memory cells and generation of antibodies, respectively.

  • Memory: Memory of previously encountered viruses and pathogens is kept in the form of memory cells, so that if the same pathogen attacks the immune system in the future, it can trigger a fast and more effective response.

  • Multi-layered: An NIS has various layers. The first layer of defense is human skin and various body secretions. Apart from that innate and adaptive immune systems are the two main layers in an NIS to provide protection against various pathogens.

  • Decentralized process: An NIS is decentralized in nature, meaning it does not have any central control.

  • Noise tolerance: An NIS is tolerant to noise, and a perfect match between pathogen and immune cell receptors is not required to trigger an immune response.

  • Dynamic system: An NIS is constantly under attack by new pathogens and therefore it is constantly changing and adapting to new pathogens. As pathogens and viruses are evolving all the time, an NIS has to be dynamic to trigger an appropriate immune response.

Complete Chapter List

Search this Book: