John E. Hunt

An artificial immune system for data analysis

We present a simplified view of those parts of the human immune system which can be used to provide the basis for a data analysis tool. The motivation for and reasoning behind such a model is given and the desire for a transparent model and meaningful visualization and interpretation techniques is noted. A minimalist formulation of an artificial immune system and some of its behaviour is described. A simple implementation and a suitable visualization technique are demonstrated using some trivial data and the famous iris data set.

Jisys: The Envelopment of an Artificial Immune System for Real World Applications

This chapter Enscribes a machine learning system based on metaphors taken from the human immune system. This learning system, known as an Artificial Immune System (AIS), has been Enveloped over the past 3 years. The current implementation,Jisys,embodies the results of this research. However, the Jisys implementation requires further Envelopment as well as application to complex real world problems. This chapter Enscribes future Envelopments ofJisysas well as consiEnration of how it can be applied to a complex problem in the domain of mortgage fraud Entection. It should not be read as a Ensign document, although it contains elements of such a document, rather it should be read as an indication of the directions which need to be followed, the issues which need to be addressed and some suggested solutions

Data analysis using artificial immune systems, cluster analysis and Kohonen networks: some comparisons

Knowledge discovery in databases (KDD) is still a relatively new and expanding field. To aid the KDD process, data mining methods are used to extract previously unknown patterns and trends in vast amounts of data. There exist a number of data mining techniques, taking methods from the machine learning, statistical analysis and pattern recognition communities, to name a few. Each technique has something different to offer over other techniques and each is suitable for different purposes giving certain benefits in varying situations. This paper examines a novel data analysis technique that is inspired by the human immune system: the artificial immune system (AIS). Immune system principles act as inspiration, allowing the creation of a network of cells that in effect clusters similar patterns and trends together. It is inspired by but not a model of the human immune system. This clustering allows the human user to effectively identify areas of similarity from the training data set that would previously have been unobtainable.

An adaptive, distributed learning system based on the immune system

The immune system enables human survival of infection and disease; when the system fails to work, or is defeated by a particular infection, human life is put at risk. As such it is one of the most important biological mechanisms humans possess. However, little attention has been paid to computer systems which use the immune system as their biological metaphor. In this paper we describe a learning system which is based on both the genetic mechanisms used to construct antibodies and on the influence of the immune network (which acts as a reinforcement memory). This unique combination results in a system which is self-organising, possesses no central controller, uses one-shot learning, possesses an explicit representation of what it has learnt and can forget little used information. This system is illustrated on a simple naughts and crosses (tic-tac-toe) application.

FAILURE MODE EFFECTS ANALYSIS: A PRACTICAL APPLICATION OF FUNCTIONAL MODELING

Knowledge of how a device works is important for many tasks. Yet, systems that attempt to base their reasoning on the use of a functional model fail to capture such knowledge or only capture it implicitly. Instead they rely solely on the knowledge of the purpose of the system and provide causal explanations of how this purpose is achieved. This type of model only represents knowledge of what the system is for, not how the system works. However, engineers also rely on knowledge of how a device works to complete tasks successfully. One such task is failure mode effects analysis (FMEA). FMEA involves investigation and assessment of the effects of all possible failure modes on a system. This process is both tedious and time consuming, and it requires detailed expert knowledge of the system under consideration, including information about the structure of the system and its purpose or function. This means that any attempt to automate the whole of the FMEA process must involve both the structural and functional...

Case Memory and Retrieval Based on the Immune System

A variety of case memory organisations and case retrieval techniques have been proposed in the literature. Each of these has different features which can affect how useful they are for different applications. However, in applications which are likely to hold very large numbers of cases, which are highly volatile, and the structure of which is poorly understood, most of the current approaches are unsuitable.

Evolutionary Case Based Design

This paper extends the basic framework of case based reasoning (CBR) to include an evolutionary approach to adaptation. Such an extension allows the CBR system to consider a number of alternatives in parallel rather than forcing it to make a choice about the most appropriate case to process or the best way to modify that case. This results in a system which integrates the efficiency benefits of a CBR system with the flexibility of an evolutionary system, to provide an evolutionary CBR system. The paper illustrates how such a system can be used to solve design problems in a simple civil engineering application.

Combining functional and structural reasoning for safety analysis of electrical designs

Increasing complexity of design in automotive electrical systems has been paralleled by increased demands for analysis of the safety and reliability aspects of those designs. Such demands can place a great burden on the engineers charged with carrying out the analysis. This paper describes how the intended functions of a circuit design can be combined with a qualitative model of the electrical circuit that fulfils the functions, and used to analyse the safety of the design. FLAME, an automated failure mode and effects analysis system based on these techniques, is described in detail. FLAME has been developed over several years, and is capable of composing an FMEA report for many different electrical subsystems. The paper also addresses the issue of how the use of functional and structural reasoning can be extended to sneak circuit analysis and fault tree analysis.

Augmenting an artificial immune network

The human immune system can provide many metaphors that can be utilised effectively in the field of machine learning. These metaphors have been successfully applied to the complex real world problem of mortgage fraud detection, using a learning system known as Jisys. The Jisys system identifies patterns in mortgage fraud data by constructing an immune network, which is then evolved through the analysis of additional fraudulent and no-fraudulent applications. By viewing this network, a human expert can gain a better understanding of the fraudulent behavior. This paper describes significant developments over the original Jisys system. For example, the network is currently a flat structure with significant groupings within it. However, it can be difficult to identify these groups, to analyse them and then determine their significance. This paper describes some advances, which significantly improve the interpretation of the network. We also consider various statistical techniques which can be used to enhance the performance of the Jisys system as well as exploiting inherent properties of the network which enable significant performance improvements to be implemented (with not loss of information content). Finally, the paper presents some analysis of the structures generated by the Jisys system and relates them back to the known structures in the training data set.

Evolving hierarchical robot behaviours

Inspired by the work of Brooks, many researchers involved in programming robots have turned to the behaviour-based approach. At present, the behaviours are designed by hand and hard-wired into the architecture. The work presented in this paper looks at using an evolutionary algorithm approach (based on the genetic algorithm) to construct behaviours. Building from well-defined primitive behaviours, hierarchies can be evolved to produce more complex behaviour. The behaviours in the evolutionary system are tested in simulation, but the best are then tested on a mobile robot for grounding in the real world. This allows the evolutionary process to rapidly drive the development of the behaviours using simulation while also ensuring their suitability in the real world. In the paper we show how this evolutionary process evolves practical hierarchical behaviours for the detection of a goal object in a series of mazes.