Recently United Kingdom research community has proposed several challenges in computer science research with the ambition for development implementation in a wide research area. These Grand challenges would address areas in Non-classical computation which would include the exploitation of the natural world such as DNA and quantum computing in order to develop new areas of computations key being the artificial immune system (AIS). This has been well commented considering the lack of certain issues key being the lack of thought in areas of AIS, theoretical work and the limited view of the immune system. However it has to be agrees that many of these areas are not new since they have been raised through AIS communities.
Artificial immune system
So as to be able to establish a common basis for AIS, a proposal based on works by de Castro and Timmis has furthered an idea of a framework for Artificial Immune system. The argument raised is that in the case of other biological inspired approaches, such as in artificial neural networks and evolutionary algorithms(as mentioned previously), such a basic idea of a framework already exists and as such can help with the understanding and construction of such systems.
For instance considering a set of artificial neurons that can be arranged to together to form an artificial neural network, we can be able to acquire knowledge by learning from the adaptive process these neural networks undergo by altering the parameters within their frameworks. As a result of this process a population of individuals can be evolved in an artificial manner.
The immune system from an artificial immune system perspective
A majority of developments within the artificial immune system is focused on three main immunological theories which are: clonal selection, immune networks and negative selection. Researchers have relied in clonal selection and immune theories as a basis for the learning and memory mechanism of the immune system. As for the identification of anomalous entities, it is the negative selection theories that have been used.
It can further be established that biological inspirations behind artificial immune (AIS) system has been held back and the mechanism as well as the processes within the immune system exploited by AIS communities have been perceived limitedly. With this in mind we should all come into one agreement that despite the successes to date AIS the restricted view of the immune system adopted by the AIS practitioners will hamper the success if AIS.
With regards to vertebrate’s immune system, it is known that it is composed of diverse sets of cells and molecules that work together with other systems such as neural and endocrines so as to maintain a steady state within the host. There are two basic types of immune system: the adaptive and the innate. The innate immunity is not always directed towards the specific invaders, but rather against general pathogens whereas the innate immune system play an important role in regulating immune responses included the adaptive immune responses. However it has been found that the immune system is by no means capable of completely protecting the body. With adaptive or acquired immunity the immune system is able to launch as attack against any invaders that the innate systems cannot rid of.
Major works on AIS revolves around the clonal selection theory. The clonal selection system is a term used to describe the basic properties of an adaptive immune response to an antigenic stimulus. It establishes the concept that those cells capable of recognizing an antigenic stimulus will proliferate and thus being selected against that which do not. For the immune system to be protective over some time, antigen recognition is not sufficient. In the normal course of immune system evolution, an organism is supposed to encounter a given antigen repeatedly during its tenure.
With regards to computations this has enabled the development of population-based algorithms inspired by clonal selection process. With a perspective from a computation point of view, it can be established that the application of clonal selection theory would lead algorithms that evolve through cloning, mutation and selection phase. This would create a solution in terms of optimization or pattern detection during learning.
Negative selection refers to a process by which selection takes place in the thymus glands. This is facilitated by T-cells that are produced in the bone marrow and undergo maturation in the thymus glands before they are released into the lymphatic system.
It is through this negative selection that works by forest and company has been inspired into the development negative selection algorithms that detect data manipulation caused by computer virus.
Building artificial immune system (AIS)
Before constructing an AIS, there are many computational and practical approaches that have to be considered, these includes: computational complexity of the approach which relates to the time and space needed to generate a suitable number of detector in other word the members of the population that are required for the job.
Secondly the aspect of data to be used has to be considered. However this would mean that one has to be careful from abstracting away from the underlying data representation in this case the real values of the sensors to ensure that there is accurate mapping between higher levels representations and the actual system.
Challenges posed in artificial immune system (AIS)
One of the key challenges faced by AIS is the development of accurate metaphors and novel that are a benefit to immunology. This is due to the use of naïve approaches of extracting metaphors from the immune system. This has led to AIS drifting away from its immunological root. However if greater interaction with mathematicians and immunologist is maintained useful models can be used as a basis for abstraction into powerful algorithms.
Secondly the fact that present works on AIS has mainly concentrated on what other paradigms do such as optimizations and learning the focus of AIS to benefit the adoption of the immune response is not clear.
Thirdly has been the challenge to develop a theoretical basis for the AIS. Due to the fact that much work on AIS has focused on simple extraction of metaphors and direct application. Despite the availability of a framework for developing AIS, there is still significant lack in formals as well as the theoretical underpinning. This is further expounded by the fact that AIS has been applied to a wide domain of problems and yet the understanding of AIS has not been fully exploited.
Finally another challenge posed by AIS is the integration of immune and other systems. Considering the fact that the immune system does not work in isolation, therefore focus should be given on the potential of the immune system not only as inspiration but also the availability of other systems that interact with the immune system. This will pose greater benefits leading towards the understanding of the roles and functions of the immune system that would lead to development of immune inspired algorithms.
In an attempt to draw occlusions on current state of AIS we have to look at it from a holistic view so as to gain an impression on how well or how not the area of AIS has been advancing. To strengthen the assurance in significant developments in the area of AIS it will be worth considering that this area has recently received a great deal of interest by people not only to critically think about how and why they can develop and apply this immune inspired ideas but the fact that they have gone a step further in developing these systems. This best provides a sense of optimism in the future research of AIS.
Timmis, J. (2007). Artificial immune systems—today and tomorrow. Nat Comput , 1–18.