Essay: immunological inspired novelty detection

This document is interesting for a number of reasons. Ultimately, this work provides a description of the space I am working in without defining the specifics of what that research will consist of. A landscape is painted that provides a broader perspective for my proposed research project indicating where the cogs that will be the focus of the work fit into the larger system and how that larger system is connected to the real-world. The selected research area is that of using inspirations from the biological immune system of vertebrate organisms as the basis for a system that addresses the constraints and requirements of novelty detection in dynamical and high data load environments. The general research question as to how this could be achieved and the effectiveness of the resulting solution is framed within clear definitions. A concise mean of biological inspired computation is provided and a powerful yet simple framework is defined. This framework permits both existing related work to be naturally integrated and related as well as providing a guide for identifying relationships and ultimately scope for research questions posed in the area of biologically inspired computation. Further very clear roles are defined for the selected biological metaphor and its relation to the selected problem domain. A problem centric view is taken where all characteristics of the proposed solution are defined in the context of the domain in this case novelty detection. The problem is explored from both a reductionist standpoint so as that it can be easily related to existing work and form a specialist standpoint so that it can be related to practical engineering problem domains. Problem constraints are clearly laid out and from those constraints useful requirements are defined of which a solution must demonstrate meeting to satisfy the selected problem domain. The two primary problem constraints of interested are high volume input data and a dynamical underlying models both obviously in the context of novelty detection. The metaphor is briefly discussed and though it is the inspiration for a proposed general solution its usefulness ends after the formulation of a solution. A general thesis is proposed as already mentioned and a general broader system is described that represents an ideal or vision for a system. The system is intentionally wide reaching and vague. It consists of a number of specific processes the aggregation of which defines the capability of the proposed solution. It is these processes which are expected to be the focus of my work though with the knowledge that each algorithm addresses a specific need within the described larger system. Finally some useful analytical and empirical techniques are identified and discusses as being potentially useful as evidence in constructed arguments defending the proposed solution. Ultimately this work is intended to provide an induction and broad encompassing understanding of the nature of the proposed research problem. It is indented to demonstrate the usefulness of the selected problem and indicate where the fruit of such work fits into the broader scope of addressing practical engineering problems using inspirations from nature. [Introduction]