The synthesis of abiotic life-like behaviour in complex chemical systems is one of the great scientific challenges in today’s research environment. Very often in this type of design, the parameter space is so large and the system so complex that analytical, rational design techniques are extremely difficult to manage, and more often than not, unavailable altogether. Machine learning methods have found many applications in the realm of design and manufacture and the research described in this thesis describes the application of these tools towards the development of pre-specified chemical functionality in complex systems. A detailed description of the ‘Evolutionary Engine’ built with this sort of design in mind is given, including preliminary investigations into coupling this engine to a ‘real life’ chemical reactor array. Studies are performed on a range of complex systems, including benchmark problems based on cellular automata, and, for the first time in this domain, on real world problems in self-organised scanning probe microscopy. Given a target behaviour of the system in question, usually represented by a series of patterns in a 2D image, it is shown that parameters can be ‘reverse engineered’ through a sophisticated combination of machine learning techniques and image analysis methods, such that the target behaviour/pattern can be faithfully reproduced. Finally, techniques for the approximation of a complex system and its associated fitness function are explored, giving rise to a dramatic decrease in computation time with little compromise to the quality of results.

Over the past few years, IRC bots, malicious programs which are remotely controlled by the attacker, have become a major threat to the Internet and its users. These bots can be used in different malicious ways such as to launch distributed denial of service (DDoS) attacks to shutdown other networks and services. New bots are implemented with extended features such as keystrokes logging, spamming, traffic sniffing, which cause serious disruption to targeted networks and users. In response to these threats, there is a growing demand for effective techniques to detect the presence of bots/botnets. Currently existing approaches detect botnets rather than individual bots. In our work we present a host-based behavioural approach for detecting bots/botnets based on correlating different activities generated by bots by monitoring function calls within a specified time window. Different correlation algorithms have been used in this work to achieve the required task. We start our work by detecting IRC bots' behaviours using a simple correlation algorithm. A more intelligent approach to understand correlating activities is also used as a major part of this work. Our intelligent algorithm is inspired by the immune system. Although the intelligent approach produces an anomaly value for the classification of processes, it generates false positive alarms if not enough data is provided. In order to solve this problem, we introduce a modified anomaly value which reduces the amount of false positives generated by the original anomaly value. We also extend our work to detect peer to peer (P2P) bots which are the upcoming threat to Internet security due to the fact that P2P bots do not have a centralized point to shutdown or traceback, thus making the detection of P2P bots a real challenge. Our evaluation shows that correlating different activities generated by IRC/P2P bots within a specified time period achieves high detection accuracy. In addition, using an intelligent correlation algorithm not only states if an anomaly is present, but it also names the culprit responsible for the anomaly.

An increasing number of devices are being situated in public spaces yet interactions with such devices are problematic: they tend to be impersonal and subject to social apprehension while devices suitable for specific tasks may be difficult to locate. This thesis considers how one might design for these environments to overcome these issues and deliver engaging user experiences. It proposes the coupling of the interactive features of mobile and situated devices to facilitate personalised interactions with those situated devices. The thesis explores coupling techniques that extend the computational capabilities of the situated device through the addition of the input, output and storage capabilities of the mobile device. Finally it considers how multiple points of coupling can be used to link sequences of interactions with different situated devices providing rich, cohesive experiences across an environment. The thesis presents a novel framework that builds upon previous work. Existing work is reviewed that links mobiles with single situated displays, and that uses mobiles for mediating exploration of physical spaces to address the lack of work addressing multiple situated de-vices in public. This review grounds a proposal and elaboration of a core model of interaction within a coupling environment, providing the basis for a design framework. This was sup-ported by the implementation of a test-bed that consisted of six couples in various configura-tions, underpinned by a software infrastructure. Formative user studies refined the framework and revealed novel aspects of the user experi-ence for study. It was found that through support for narrative and personal orchestration, coupling environments afford personalised trajectories. By designing for personal trajectories the visitor has a more enjoyable personal experience and seeks to improve the experiences of others. In addition, coupling environments support social experiences; the step-by-step nature of a visitor’s trajectory through the coupling environment lends itself to gradually introducing visitors to social coupled interaction and reducing social awkwardness.

This thesis is concerned with single and multiple target visual tracking algorithms and their application in the real world. While they are both powerful and general, one of the main challenges of tracking using particle filter-based algorithms is to manage the particle spread. Too wide a spread leads to dispersal of particles onto clutter, but limited spread may lead to difficulty when fast-moving objects and/or high-speed camera motion throw trackers away from their target(s). This thesis addresses the particle spread management problem. Three novel tracking algorithms are presented, each of which combines particle filtering and Kernel Mean Shift methods to produce more robust and accurate tracking. The first single target tracking algorithm, the Structured Octal Kernel Filter (SOK), combines Mean Shift (Comaniciu et al 2003) and Condensation (Isard and Blake 1998a). The spread of the particle set is handled by structurally placing the particles around the object, using eight particles arranged to cover the maximum area. Mean Shift is then applied to each particle to seek the global maxima. In effect, SOK uses intelligent switching between Mean Shift and particle filtering based on a confidence level. Though effective, it requires a threshold to be set and performs a somewhat inflexible search. The second single target tracking algorithm, the Kernel Annealed Mean Shift tracker (KAMS), uses an annealed particle filter (Deutscher et al 2000), but introduces a Mean Shift step to control particle spread. As a result, higher accuracy and robustness are achieved using fewer particles and annealing levels. Finally, KAMS is extended to create a multi-object tracking algorithm (MKAMS) by introducing an interaction filter to handle object collisions and occlusions. All three algorithms are compared experimentally with existing single/multiple object tracking algorithms. The evaluation procedure compares competing algorithms' robustness, accuracy and computational cost using both numerical measures and a novel application of McNemar's statistic. Results are presented on a wide variety of artificial and real image sequences.

This thesis presents a comparison between two simulation methods, namely Discrete Event Simulation (DES) and Agent Based Simulation (ABS). In our literature review we identified a gap in comparing the applicability of these methods to modelling human centric service systems. Hence, we have focused our research on reactive and different level of detail of proactive of human behaviour in service systems. The aim of the thesis is to establish a comparison for modelling human reactive and different level of detail of proactive behaviour in service systems using DES and ABS. To achieve this we investigate both the similarities and differences between model results performance and the similarities and differences in model difficulty performance. The comparison of the simulation methods is achieved by using a case study approach. We have conducted three case studies, the choice of our case study systems taking into consideration the number of different key proactive behaviours that can be observed. In the first case study (fitting room services) we consider single proactive staff behaviour, in the second case study (international support services) we consider two proactive staff behaviours and, finally, the third case study (airline check-in services) considers three proactive staff behaviours. The proactive behaviours considered are: taking charge from experience, taking the initiative to fulfil a goal and supervising by learning. To conduct our case studies we have created two sets of simulation models. The first set consists of one DES model for each of the case studies. As service systems have an organisational structure we could not implement our agent-based simulation models purely as agent-based models. Instead, for the second set we have created combined DES/ABS models (one for each case study), where the DES part represents the system and the ABS part represents the active entities inside the system (i.e. the people).With these models we have carried out two sets of experiments: Set A is concerned with modelling results performance, while set B is related to model difficulty performance. We have then conducted statistical analysis on the results of these experiments. Evidence from the experiments reveals that DES and combined DES/ABS are found suitable to model the reactive and most levels of proactive behaviour modelled in this thesis. In addition, combined DES/ABS is found more suitable for modelling higher levels of proactive behaviour (complex behaviour). Another finding from the experiments is that it is only worth representing complex proactive behaviour if it occurs frequently in the real system (considering the relation between modelling effort and impact). The contribution made by this thesis to the body of knowledge is the comparison of DES and combined DES/ABS for modelling human reactive and different level of detail of human proactive behaviour in service systems. This comparison will assist modellers who are new to the field of service systems modelling to make an informed decision on the method they should use for their own modelling, based on the level of proactiveness inherent in the real system and on the levels of difficulties they should expect for each method.

Algorithmic problem solving provides a radically new way of approaching and solving problems in general by using the advances that have been made in the basic principles of correct-by-construction algorithm design. The aim of this thesis is to provide educational material that shows how these advances can be used to support the teaching of mathematics and computing. We rewrite material on elementary number theory and we show how the focus on the algorithmic content of the theory allows the systematisation of existing proofs and, more importantly, the construction of new knowledge in a practical and elegant way. For example, based on Euclid’s algorithm, we derive a new and efficient algorithm to enumerate the positive rational numbers in two different ways, and we develop a new and constructive proof of the two-squares theorem. Because the teaching of any subject can only be effective if the teacher has access to abundant and sufficiently varied educational material, we also include a catalogue of teaching scenarios. Teaching scenarios are fully worked out solutions to algorithmic problems together with detailed guidelines on the principles captured by the problem, how the problem is tackled, and how it is solved. Most of the scenarios have a recreational flavour and are designed to promote self-discovery by the students. Based on the material developed, we are convinced that goal-oriented, calculational algorithmic skills can be used to enrich and reinvigorate the teaching of mathematics and computing.

With the advent of digital printing presses and the continued development of associated technologies, variable data printing (VDP) is becoming more and more common. VDP allows for a series of data instances to be bound to a single template document in order to produce a set of result document instances, each customized depending upon the data provided. As it gradually enters the mainstream of digital publishing there is a need for appropriate and powerful editing tools suitable for use by creative professionals. This thesis investigates the problem of representing variable data documents in an editable visual form, and focuses on the technical issues involved with supporting such an editing model. Using a document processing model where the document is produced from a data set and an appropriate programmatic transform, this thesis considers an interactive editor developed to allow visual manipulation of the result documents. It shows how the speed of the reprocessing necessary in such an interactive editing scenario can be increased by selectively re-evaluating only the required parts of the transformation, including how these pieces of the transformation can be identified and subsequently re-executed. The techniques described are demonstrated using a simplified document processing model that closely resembles variable data document frameworks. A workable editor is also presented that builds on this processing model and illustrates its advantages. Finally, an analysis of the performance of the proposed framework is undertaken including a comparison to a standard processing pipeline.

The field of physical modelling and simulation plays a vital role in advancing numerous scientific and engineering disciplines. To cope with the increasing size and complexity of physical models, a number of modelling and simulation languages have been developed. These languages can be divided into two broad categories: causal and noncausal. Causal languages express a system model in terms of directed equations. In contrast, a noncausal model is formulated in terms of undirected equations. The fact that the causality can be left implicit makes noncausal languages more declarative and noncausal models more reusable. These are considered to be crucial advantages in many physical domains. Current, mainstream noncausal languages do not treat equational models as first-class values; that is, a model cannot be parametrised on other models or generated at simulation runtime. This results in very limited higher-order and structurally dynamic modelling capabilities, and limits the expressiveness and applicability of noncausal languages. This thesis is about a novel approach to the design and implementation of noncausal languages with first-class models supporting higher-order and structurally dynamic modelling. In particular, the thesis presents a language that enables: (1) higher-order modelling capabilities by embedding noncausal models as first-class entities into a functional programming language and (2) efficient simulation of noncausal models that are generated at simulation runtime by runtime symbolic processing and just-in-time compilation. These language design and implementation approaches can be applied to other noncausal languages. This thesis provides a self-contained reference for such an undertaking by defining the language semantics formally and providing an in-depth description of the implementation. The language provides noncausal modelling and simulation capabilities that go beyond the state of the art, as backed up by a range of examples presented in the thesis, and represents a significant progress in the field of physical modelling and simulation.

Fuzzy rule-based systems (FRBS) use the principle of fuzzy sets and fuzzy logic to describe vague and imprecise statements and provide a facility to express the behaviours of the system with a human-understandable language. Fuzzy information, once defined by a fuzzy system, is fixed regardless of the circumstances and therefore makes it very difficult to capture the effect of context on the meaning of the fuzzy terms. While efforts have been made to integrate contextual information into the representation of fuzzy sets, it remains the case that often the context model is very restrictive and/or problem specific. The work reported in this thesis is our attempt to create a practical frame work to integrate contextual information into the representation of fuzzy sets so as to improve the interpretability as well as the accuracy of the fuzzy system. Throughout this thesis, we have looked at the capability of the proposed context dependent fuzzy sets as a stand alone as well as in combination with other methods in various application scenarios ranging from time series forecasting to complicated car racing control systems. In all of the applications, the highly competitive performance nature of our approach has proven its effectiveness and efficiency compared with existing techniques in the literature.

Self-assembly is a ubiquitous process in nature in which a disordered set of components autonomously assemble into a complex and more ordered structure. Components interact with each other without the presence of central control or external intervention. Self-assembly is a rapidly growing research topic and has been studied in various domains including nano-science and technology, robotics, micro-electro-mechanical systems, etc. Software self-assembly, on the other hand, has been lacking in research efforts. In this research, I introduced Automated Self-Assembly Programming Paradigm (ASAP²), a software self-assembly system whereby a set of human made components are collected in a software repository and later integrated through self-assembly into a specific software architecture. The goal of this research is to push the understanding of software self-assembly and investigate if it can complement current automatic programming approaches such as Genetic Programming. The research begins by studying the behaviour of unguided software self-assembly, a process loosely inspired by ideal gases. The effect of the externally defined environmental parameters are then examined against the diversity of the assembled programs and the time needed for the system to reach its equilibrium. These analysis on software self-assembly then leads to a further investigation by using a particle swarm optimization based embodiment for ASAP². In addition, a family of network structures is studied to examine how various network properties affect the course and result of software self-assembly. The thesis ends by examining software self-assembly far from equilibrium, embedded in assorted network structures. The main contributions of this thesis are: (1) a literature review on various approaches to the design of self-assembly systems, as well as some popular automatic programming approaches such as Genetic Programming; (2) a software self-assembly model in which software components move and interact with each other and eventually autonomously assemble into programs. This self-assembly process is an entirely new approach to automatic programming; (3) a detailed investigation on how the process and results of software self-assembly can be affected. This is tackled by deploying a variety of embodiments as well as a range of externally defined environmental variables. To the best of my knowledge, this is the first study on software self-assembly.