Ph.D. / Many organisations are starting to make large parts of their information resources publicly accessible. For example, many organisations publish information using the Internet. Some organisations allow non-employees to connect to their systems and retrieve information - many banks allow customers to retrieve account statements via the Internet. There is a trend towards more open information systems and more distributed processing such as client/server processing. The above are just some of the trends in computer information processing that creates new and complex problems in providing information systems that are both secure and manageable. To add to the complexity of the information security problem, organisations use the Internet to conduct some of their business and use many different applications, each with its own unique access control mechanisms. Central management of information security in a heterogeneous and distributed environments, such as the Internet has become a nightmare. There is a need for an information security model that will allow organisations to make use of the new trends in information processing, but still have confidence that they have adequate security and that the management of their information security systems is fairly easy. In this thesis we propose a model that satisfies the above requirements. We call this model the Intelligent Security Agent Model (ISAM). The ISAM model is based on two technologies: intelligent software agents and distributed objects. The main component of the model is Intelligent Security Agents that act as security brokers for its users in a distributed environment. In chapter 2 of the thesis, we design an Intelligent Security Agent which provides various information security services in open client/server environments. The Intelligent Security Agent Model addresses certain problems/requirements, such as single sign-on, in information security. These problems and possible solutions are described in chapter 4 to chapter 10 of this thesis. An Intelligent Security Agent must be protected from unauthorised modification, theft, etc. Chapter 3 shows how an Intelligent Security Agent is secured by implementing it as a distributed object. We show that the combination of intelligent software agents and distributed objects creates an agent that was not possible before, and solves many information security problems.In short, this thesis documents the results of a study in computer information security. The result of the study is a new information security model in which intelligent software agents and distributed objects are combined to create a security agent which acts on behalf of a user in open environments such as client/server systems and the Internet. The agent provides a set of services to its user and handles all information security related requests on behalf of its user.

M.Ing. (Electrical and Electronic Engineering) / Dissolved Gas-in-oil analysis (DGA) is used to monitor the condition of bushings on large power transformers. There are different techniques used in determining the conditions from the data collected, but in this work the Artificial Intelligence techniques are investigated. This work investigates which gases in DGA are related to each other and which ones are important for making decisions. When the related and crucial gases are determined, the other gases are discarded thereby reducing the number of attributes in DGA. Hence a further investigation is done to see how these new datasets influence the performance of the classifiers used to classify the DGA of full attributes. The classifiers used in these experiments were Backpropagation Neural Networks (BPNN) and Support Vector Machines (SVM) whereas the Principal Component Analysis (PCA), Rough Set (RS), Incremental Granular Ranking (GR++) and Decision Trees (DT) were used to reduce the attributes of the dataset. The parameters used when training the BPNN and SVM classifiers are kept fixed to create a controlled test environment when investigating the effects of reducing the number of gases. This work further introduced a new classifier that can handle high dimension dataset and noisy dataset, Rough Neural Network (RNN). This classifier was tested when trained using the full dataset and how it is affected by reducing the number of gases used to train it. The results in these experiments showed that ethane and total combustible gases attributes are core attributes chosen by the four algorithms as gases needed for decision making. The average results of the classification performance showed that the reduction of attributes helps improve the performance of classifiers. Hence the science of transformer condition monitoring can be derived from studying the relations and patterns created by the different gases attributes in DGA. This statement is supported by the classification improvements where the RNN classifier had 99.7% classification accuracy when trained using the three attributes determined by the PCA.

M.Sc. (Information Technology) / The dissertation provides a look at artificial intelligence techniques that have been embedded into computer games, which is then referred to as either game agents or non-player character agents (NPC-agent). This dissertation focuses on the design and creation of an agent that utilises the concept of symbiosis for controlling a non-player character (NPC). The dissertation considers what agents are, and what their connection to computer games is. The aspect of cloud computing was considered as it provides a new avenue for additional resources to be obtained for agents in computer games. To understand how this can be achieved cloud computing is investigated to provide a basic understanding of cloud computing as a whole. The next stage was to consider the essential components in a game agent, which are navigation, decision-making, and perception. The components are then discussed along with the techniques that they have utilised. Once the research was completed, the model was designed with the aid of symbiosis to create a symbiotic game agent (SGA). The SGA is created by breaking down the vital components of a game agent into separate symbiont agents that have the sole purpose of fulfilling one set of responsibilities, for example, a symbiont agent is responsible for all perception activity in connection to the NPC. To allow access to the resources to be obtained from cloud computing, service orientation and a service agent are introduced into the design. The last stage of the dissertation is the creation of a prototype SGA to provide proof of the concept of the model. Within the prototype, the SGA was implemented along with the implemented navigation symbiont agent, using a service agent. Results of the implementation are presented, highlighting the use of each of the symbiont agents in the SGA while they direct the NPC in the prototype computer game environment.

D.Litt. et Phil. / It is said that computer security is like getting into bed with an elephant. You know you have a problem, but you can't get your arms around it! Looking at security from a distributed point of view makes this elephant seem much bigger! The growth of the Internet (see chapter 2) is also not very comforting to computer security specialists. Companies want to start utilising the Internet for their business transactions, while the man on the street wants to use it for what they deem necessary or convenient. It is becoming more and more common placed to buy items from virtual storefronts by making use of the Internet. [17,20,23] Literature, as well as the World Wide Web has supplied us with information [34] regarding the "war" between the IT Security Professional and the hacker community. It is quite surprising to see the large number of hacker sites on the Internet [15,16,21] that publish information regarding hacked sites, as well as tools and techniques that can assist almost anyone in accomplishing some of these sometimes, daring feats. If this information is studied and if we keep in mind that some of the more serious hacking attempts are being kept secret for reasons like loss of business or credibility, it might be deduced that the hacker community is always a step or two ahead of Security Professionals. It is the purpose of this thesis to present a model that will ensure secure, anonymous communication across the Internet This model is not aimed at replacing current technologies. It merely attempts to provide an alternative method for safe communication across public networks, like the Internet. The model will make use of a number of existing technologies in conjunction with one another to achieve its goal of secure, anonymous communication. The technologies that will be used and how will be discussed briefly in this chapter.

M.Com. (Information Technology) / ADLOA is an Architecture Description Language (ADL) proposed to describe biologicallyinspired complex adaptive architectures such as ontogenetic architectures. The need for an ontogenetic ADL stems from the lack of support from existing ADLs. This dissertation further investigates the similarities between existing intelligent architectures and ontogenetic architectures. The research conducted on current ADLs, artificial ontogeny and intelligent architectures reveals that there are similarities between ontogenetic architectures and other intelligent architectures. However, the dynamism of artificial ontogeny indicates a lack of support for architecture description. Therefore, the dissertation proposes two core mechanisms to address ontogenetic architecture description. Firstly, the ADLOA process is defined as a systematisation of artificial ontogeny. The process specifies a uniform approach to defining ontogenetic architectures. Secondly, a demonstration of the implemented ADLOA process is used, in conjunction with the ADLOA model, mechanisms and Graphical User Interface (GUI), to present a workable description environment for software architects. The result of the dissertation is a standalone ADL that has the ability to describe ontogenetic architectures and to produce language-dependent code frameworks using the Extensible Markup Language (XML) and Microsoft Visual Studio platform.

The evolutionary nature of humans requires agent systems to be continuously replaced due to their inability to meet or adapt to our changing needs. Therefore, to eliminate the need for a human to continuously adapt an agent, evolutionary agents are required [Chu04, Ore99, Rak02, Syc96]. This dissertation develops a feasible option to ensuring that agents continuously develop desirable behaviour. The solution is a specialized architecture that embeds self-evolvement into a target agent. The specialized architecture ensures that desirable behaviour emerges from any agent, as it is embedded between the target agent and the target agent’s environment and therefore is able to obtain domain- and hardwarespecific information from the target agent. The specialized architecture is a comprehensive methodology that incorporates all agents with the ability to embed the required self-evolvement enhancements as domain- and hardwarespecific information is obtained from the target agent. The specialized architecture responsible for embedding self-evolvement into an agent is the generic self-evolvement effecting evolutionary agent (GSEEA). The GSEEA is developed with a single goal, which is to ensure that the target agent meets the requirements of a changing environment. Changing environmental conditions can include different network conditions and different platforms. The GSEEA’s goal is accomplished by embedding the required self-evolvement enhancements into the target agent to produce a self-evolvement enhanced agent. In this dissertation the GSEEA is implemented to demonstrate its feasibility and problem-solving accuracy. In the GSEEA implementation the target agent is a puzzle-solving agent and the self-evolvement enhanced agent is the selfevolvement enhanced puzzle-solving agent. The GSEEA’s deliberative component consists of two algorithms, namely a genetic algorithm and a learning algorithm. The GSEEA’s genetic algorithm develops knowledge base rules (selfContents III evolvement enhancements) that modify actuator information. The GSEEA’s learning algorithm updates developed knowledge base rules by modifying sensor information. The GSEEA tests the developed self-evolvement enhancements by embedding them into the target agent through the target agent’s knowledge base manager, evaluating the developed self-evolvement enhancements and deleting those which do not enhance the target agent. The target agent achieves selfevolvement as additional enhancements required by the self-evolvement enhanced agent can be achieved by applying the same process followed to enhance the target agent which was discussed previously. The evaluation of the GSEEA implementation demonstrated that the GSEEA was implemented successfully based on feasibility and problem-solving accuracy as the self-evolvement enhanced puzzle-solver agent outperformed the puzzlesolver agent. / Prof. E.M. Ehlers

Ehlers, E.M., Prof.

This thesis reports on research into the applicability of intelligent agents in the airline scheduling environment. The methodology employed was to look at intelligent agent research and then, based on this, to build models that can be used to solve some of the airline scheduling problems. The following was done: · An agent-based model was developed that can assist airline schedulers in the maintenance of a disrupted schedule. The agent model consists of a hybrid approach combining elements of machine learning and expert systems. · A multiagent model was developed that can generate a profitable and flyable schedule. The multiagent model developed extends the traditional control structures of the hierarchical agent organisation to a matrix structure. This new model can be extended to any problem domain that deals with resource allocation and capacity management. To guide the thinking behind this research, a few questions were posed regarding the problem to be solved: Q1. Can intelligent agents play a role in the airline industry, with specific focus on the scheduling creation and maintenance process? Q2. What will the design of the agent models be if the scheduling needs of an airline have to be addressed? Q3. If the models as envisioned in question 2 can be created, what will the practical implications be? At a conceptual level the research produced three results: R1. No references were found to multiagent technology in the production or maintenance of airline schedules. This theoretical research into agent systems shows that there is applicability in the scheduling environment, with specific reference to schedule maintenance and generation. R2. An agent model was created that combines declarative knowledge with empirical learning to assist human schedulers in the day-to-day maintenanceof the schedule. Multiple solutions to a scheduling problem are generated by the agent using embedded scheduling rules. The agent then uses the Qlearning algorithm to learn the preferences of the human scheduler. This approach combines the best of expert systems and machine learning. To solve the problem of schedule generation, a multiagent system with a matrix governance model was introduced. Aircraft and airports were modelled as buying and selling agents. The business manager agent that assigns individual aircrafts to specific routes was defined. This was accomplished by matching individual aircraft capacity to origin-destination demand. The agent model was then expanded to show how the inclusion of a resource manager agent can handle system capacity management. This is a matrix governance model, as an aircraft agent is managed by a business manager agent, as well as by a resource manager agent. The initial results from the prototype show that this model can generate profitable and flyable schedules. The multiagent model developed extends the traditional hierarchical agent organisation to that of a matrix structure. The contract net protocol used for typical multiagent coordination was adapted to work in this new control structure. This new model can be extended to any problem domain that deals with resource allocation and capacity management. R3. A few airlines use expert systems to handle schedule disruptions. By introducing machine learning, a flexibility is achieved that is currently not available. The approach proposed for schedule generation is not guaranteed to provide optimal results like traditional operations research techniques, but it is useful for high-level analysis, long-term planning, new hub or alliance planning and research. It also has potential as a catalyst for integrated planning. Keywords: Multiagent systems, airline scheduling / Ehlers, E.M., Prof.

M.Sc. (Computer Science) / General machine intelligence represents the principal ambition of artificial intelligence research: creating machines that readily adapt to their environment. Machine learning represents the driving force of adaptation in artificial intelligence. However, two pertinent dilemmas emerge from research into machine learning. Firstly, how do intelligent agents learn effectively in real-world environments, in which randomness, perceptual aliasing and dynamics complicate learning algorithms? Secondly, how can intelligent agents exchange knowledge and learn from one another without introducing mathematical anomalies that might impede on the effectiveness of the applied learning algorithms? In a robotic search and rescue scenario, for example, the control system of each robot must learn from its surroundings in a fast-changing and unpredictable environment while at the same time sharing its learned information with others. In well-understood problems, an intelligent agent that is capable of solving task-specific problems will suffice. The challenge behind complex environments comes from fact that agents must solve arbitrary problems (Kaelbling et al. 1996; Ryan 2008). General problem-solving abilities are hence necessary for intelligent agents in complex environments, such as robotic applications. Although specialized machine learning techniques and cognitive hierarchical planning and learning may be a suitable solution for general problem-solving, such techniques have not been extensively explored in the context of cooperative multi-agent learning. In particular, to the knowledge of the author, no cognitive architecture has been designed which can support knowledge-sharing or self-organisation in cooperative multi-agent learning systems. It is therefore social learning in real-world applications that forms the basis of the research presented in this dissertation. This research aims to develop a distributed cognitive architecture for cooperative multi-agent learning in complex environments. The proposed Multi-agent Learning through Distributed Adaptive Contextualization Distributed Cognitive Architecture for Multi-agent Learning (MALDAC) Architecture comprises a self-organising multi-agent system to address the communication constraints that the physical hardware imposes on the system. The individual agents of the system implement their own cognitive learning architecture. The proposed Context-based Adaptive Empathy-deliberation Agent (CAEDA) Architecture investigates the applicability of emotion, ‘consciousness’, embodiment and sociability in cognitive architecture design. Cloud computing is proposed as a method of service delivery for the learning system, in which the MALDAC Architecture governs multiple CAEDA-based agents. An implementation of the proposed architecture is applied to a simulated multi-robot system to best emulate real-world complexities. Analyses indicate favourable results for the cooperative learning capabilities of the proposed MALDAC and CAEDA architectures.

No description available.