Speaker	Title and Abstract	Date and Location
Dr. Andy Greensted	Computer crashes; is it all the software engineers fault? - An introduction to hardware fault tolerance. In an age when the maturity of electronic components gives a feeling of guaranteed dependable operation of everyday electronic hardware items, is the application of complex reliability engineering techniques still necessary? The aim of this seminar is to provide a basic introduction to the type of faults that still plague electronic hardware, and the techniques, both traditional and novel, used to help tolerate them.	21st Oct. 2005. CS103
Heather Turner	Rule Migration: Can we design/ emergence in a cellular-automaton-like system? If a required behaviour is designed into a system, then many would argue that this is not emergent behaviour. Rule migration allows system behaviour to be specified and designed at the level of the emergence, the translates that behaviour into rules for a lower-level model, like a cellular automata (CA). The high level behaviour will be emergent relative to the CA rules. This talk will provide a breif background to both emergence, and cellular automata, before describing how rule migration works, using a simple exmaple.	4th Nov. 2005. P/X/001
Dr. Dave Chesmore	Automated Identification of Species: Correct identification of species is vitally important in many fields, particularly for economically important species (e.g. plant dieases, fungi, insects). Automated species identification (ASI) is now reaching maturity as a research theme. The seminar will describe the issues surrounding ASI and techniques employed in York for bioacoustic applications (e.g. bats, wood-boring beetle larvae) and image-based applications (quarantine fungal pathogens, insect identification). These applications involve various forms of sensor technology and AI including MLPs, syntactic PR and SOMs.	18th Nov. 2005. CS103
Dr. Kate Jeffery (UCL)	How does a rat know where it is?: The question of how the brain represents 'place' has been of enormous interest to behavioural neuroscientists ever since the discovery, by John O'Keefe, that neurons in the rat hippocampus become active when the animal enters a particualr part of its environment. The so-called 'place cells' are generally important becasue it seems that if we can understand how they know where they are, we can understand how the rat knows where it is. My research has focused on how place cells integrate information of various different functional classes - directional, geometric and 'contextual' - to decide whether to fire or not. It appears that different kinds of information are processed differently - directional cues orient the representation as a whole, gemetric cues are integrated across two (and possibly three) dimensions to produce focal firing locations, and contextual cues act like a switch to tell the place cell whether to express a particular field in that place on that occasion. Collectively, it seems that the place cells can represent not only the place but also the 'context', and thus may form the substrate for the formation of episodic memory.	2nd Dec. 2005. P/T/111
Christmas Break
James Walker	The Automatic Acquisition, Evolution and Re-use of Modules in Caretesian Genetic Programming Embedded Cartesian Genetic Programming (ECGP) is an extension of the directed graph based Cartesian Genetic Programming (CGP), which allows the automatic acquisition, evolution and re-use of modules (a form of Automatically Defined Function). ECGP is tested on various digital circuit problems and the computational effort figures are compared with the non-modular CGP. The results show that ECGP shows a substantial improvement in performance over CGP and that the computational speedup is more pronounced on larger problems. Analysis of the evolved modules shows that they of ten contain functions representing partial solutions. Prospects for further improvements to ECGP and some of the problems encountered are discussed.	20th Jan. 2006. CS103
Dr. Martyn Amos (Exeter)	The Ants and the Airbus: Agent-Based Modelling of Complex Systems: Agent-based modelling is a successful strategy for the study of many different complex systems. By taking a "bottom-up" approach to the analysis of such systems, simulation studies often capture surprising and novel aspects of their behaviour. We present a brief introduction to the agent-based paradigm, before presenting our recent work on modelling both ant colonies and aircraft evacuations. This will be a general talk, and no background is assumed.	3rd Feb. 2006. P/T/006
Dr. Mic Lones	Evolvability: The Missing Link?: Make a population, apply selective pressure, induce random changes, and repeat until the problem is solved. Meanwhile, go off and have a cup of tea. This Darwinian process is the essence of evolutionary computation, but is there more to life than tea the variation-selection paradigm? This seminar is an introduction to the concept of evolvability: the idea that a system can be organised in such a way that evolution is more likely to explore productive areas of its search space than if the system were organised differently. In particular, this seminar aims to address two questions: What is it about the organisation of biological systems that makes them evolvable? How can this knowledge be used to improve artificial evolution? Drinking of tea is optional, but encouraged. PDF of talk.	17th Feb. 2006. CS103
Pete Mendham	Agents, Autonomy and the Final Frontier Welcome to the 21st Century. We have a space launch vehicle which behaves like an aeroplane, its weekly service flight to low earth orbit delivering and collecting spacecraft. And all this without the need for human intervention during the mission, whatever goes wrong. How is this astonishing feat possible? Well, at the moment, it isn't. But we're working on it. This talk will introduce research that is being conducted in the Intelligent Systems Group concerning the Skylon spaceplane, which is required to operate autonomously. The focus of my studies has been the application of a multi-agent software architecture to the problem of controlling the vehicle and all of its functions. I will set out the reasons for this apparently insane choice, the potential advantages of using a multi-agent system, and the many problems it causes. My presentation will examine the possible forms that a multi-agent control system could take and the architectural choices that must be made to obtain the system properties that we want so badly. Covering topics as diverse as "what is autonomy?" and "what is the role of a control system?", right the way to "why can't we get good tea and coffee in Electronics?", all are welcome.	3rd March 2006. PT/006
Fiona Polack	Engineering Emergence: Manufactured complex emergent systems have been postulated for a quarter of a century, but advances in engineering focuses on overcoming hardware and scale issues. Softer aspects of engineering (eg. for agent systems) are largely applying current approaches to components, not addressing the whole system. With Susan Stepney and others, I have started to explore the needs to a new engineering discipline for complex emergent systems. Based on best practice in critical systems development, I would like the engineering to focus on procuding evidence of performance and quality - assurance-based or evidence-based engineering. I will outline where the research is (not very far from the start), and invite discussion of where it should go next. PPT of talk.	18th August, 2006. 1215-1315 2006 CS122
Cristina Santini and Andy Tyrrell	Understanding and Harnessing Self-Organisation to Engineering: PDF of abstract and references	18th August, 2006. 1315-1415 CS122
Kiran Fernandes	In this theory development paper, we synthesize and apply population ecology and configuration views to the study of manufacturing configuration formation and preservation. This involves using the competitors, stress tolerators and ruderals (CSR) model (Grime 1974) to represent a typological space of strategic archetypes that varies with stress intensity (limited resources due to supply or adoption constraints) and disturbance intensity (the destruction of firm density). Using classification data that infers and organizes manufacturing configurations for the automotive assembly sector, we examine how configuration diversity varies according to these environmental dimensions. Together the CSR model and the manufacturing configuration data provide two levels of abstraction for organizing, reviewing and generating insights about the formation and preservation of strategic groups. This provides a basis for a number of preliminary propositions concerning the congruence between configuration capability mix and the blend of stress and disturbance intensity. The paper concludes by discussing the implications of this stress-disturbance framework for managerial practice and outlines potential areas of future research.	1st Sept. 2006 1215-1315 CS122