Cybernetics Society Report

Citation

(2000), "Cybernetics Society Report", Kybernetes, Vol. 29 No. 1. https://doi.org/10.1108/k.2000.06729aab.004

Cybernetics Society Report

Cybernetics Society Report

CybCon '99-Annual Conference of The Cybernetics Society

The Cybernetics Society held its 25th Annual Conference at King's College London on 25 September 1999. A full day of papers and discussions included the following:

Martin Smith, Head of the Mobile Robots Research Unit at the University of East London, and currently Chairman of The Cybernetics Society, talked about the development of an algorithm for self-steering vision-based vehicles in visually cluttered environments.

He described some recent work in developing a fast algorithm for recognising lines of perspective and vanishing points in rooms and corridors. The purpose is to generate steering signals for mobile robots to achieve self-guidance in such environments. He began with an analysis of the problems that must be overcome for a computer to recognise useful features in a television image corrupted by noise, and described some of the common methods of achieving the recognition of lines in images. He then described a method, developed by himself and a PhD research student, Nic Snailum, to search an image for lines of perspective, rapidly and reliably, using a lap-top computer (a laptop is used to keep the cost low, the volume small, power consumption low, and to enable battery power to be used). He gave some results of experiments on the algorithm being used to guide a mobile robot in a visually cluttered, real environment.

Dr Alex Andrew, formerly of the Department of Cybernetics, University of Reading, addressed the question "Is intelligence fractal?".

Much attention has been given to possibilities for self-organisation in artificial systems such as neural nets, and at one time it was widely believed that the route to something, to be described as artificial intelligence, was through the devising of self-organising artificial neural nets. However, although artificial neural nets are now proving to be useful in many areas of application, none of them are endowed with self-organising capabilities that could let them evolve spontaneously, so as to perform the symbolic manipulations that are the concern of mainstream AI. The essentially-continuous process-control type of task, to which artificial neural nets are readily applicable, does not merge readily into the symbolic type - though presumably some such transition has occurred in biological evolution. The transition can be represented as a drastic change of topology or of similarity space and Dr Andrew argued that the study of such transitions will be an important aspect of future developments in machine learning. A curious, and probably significant, aspect of the transition is that essentially-continuous values come to be associated with symbolic processing, and these are readily treated as though they were variables of the elementary process-control type of task. Examples are the estimation of "distance from a solution" in symbolic problem solving, and Minsky's "heuristic connection" between problems. Others can be illustrated by reference to game-playing and to back-propagation of errors. Principles that were effective in elementary processing reappear in the context of symbolic manipulation, operating in the latter case in terms of derived variables of the type of "heuristic connection". It may be that the above-mentioned evolutionary, or self-organising, processes will only achieve the transition to symbolic processing if they allow the elementary organisation to play the role of blueprint or exemplar, as well as the obvious one of providing a substrate. Such a dual role, in determination of physical form, must be the basis of fractal growth. Fractal growth produces lifelike forms, and seems to be a feature of many natural systems. It can be argued that intelligence has analogous characteristics.

Dr Bernard Scott, of the Centre for Educational Technology and Development, Department of Learning Technologies, De Montfort University, talked about the background to his essay "Cybernetic explanation and development". (This was the Highly Commended runner-up in The Cybernetics Society's £1,000 Essay Prize competition, which was part of the society's recent thirtieth anniversary celebrations.)

He explained that the bulk of the essay was written in 1983, when he was studying at the University of London's Institute of Higher Education in order to become an educational psychologist. Professor C.B. Hindley read and marked the essay and encouraged him to seek to have it published. However, he wasn't entirely happy with it as it stood, and promised himself that one day he would rework it. He has finally kept that promise. His intention in the essay was to address conceptual and methodological issues in the domain of child development, looking to the philosophy of science in order to do so from first principles. He already had a background in cybernetics (and had completed a PhD in that subject at Brunel University in 1976). In his reading, he discovered that cybernetic and system-theoretic concepts were, at that time, just beginning to have some influence in both developmental studies and the philosophy of science. The essay then evolved into being a more ambitious discussion of the nature of "cybernetic explanation" and a reporting of recent developments, in cybernetics and systems theory, relevant to studies of child development. In essence, the essay has the form of a newsletter, and any originality resides in the bringing together of "news items" from a range of sources. He was conscious that his attempts at synthesis are not fully adequate, but hopes he has achieved something more than a mere pot-pourri.

Dr D.J. Stewart, formerly of the Department of Cybernetics, Brunel University, spoke about "The structure of working organisations".

In previous papers (Stewart, 1989, 1999a, 1999b) he had proposed a new, ternary concept of mechanism, which forms the basis of a new paradigm for purpose and directiveness. One practical outcome of this is an analytic technique - ternary analysis. This among other things enables any particular kind of human work to be analysed into one or more of three components. One thing that ternary analysis does is to clarify the place of evaluation and judgement in work. Another is to suggest the stages to be followed in the design and construction of work and tasks, to arrive at ternary mechanisms that will achieve the desired results (Stewart, 1999c). It also offers a way of designing human institutions to use the human beings, composing them more appropriately. It shows how to design a working environment that is divided horizontally into departments, each of which is concerned with a particular dimension of imparity. It also shows how a law of sum of intervention ratios operates in the vertical structure of such a working environment, and indicates where we should expect to be able to fit computer-based systems into human institutions, and where the human beings are best placed (Stewart, 1999d).

Dr Brian Warburton, formerly of the School of Pharmacy, University of London, a past chairman of The Cybernetics Society and currently its secretary, talked about "A predator-prey model for the global economy: control of amplitude and frequency of the components".

In two previous publications (Warburton, 1995, 1996), he had put forward the idea that it might be possible to formulate a global financial model based on the so-called "predator-prey kinetic model". This, if successfully implemented, would lead to a very stable world economy, where every nation should be able to flourish. A year later, quite independently and far more persuasively, Bernice Cohen had published a book entitled The Edge of Chaos (Cohen, 1997). She had also given a paper on the same theme at an IAC congress at Namur (Cohen, 1998). These publications asserted that the course of financial crashes in human endeavour is a direct symptom of the underlying mathematical nature of chaos in global finance. A practical question, which has frequently been asked since, is: "Which pressure groups would have any conceivable interest in helping to back such a radical idea?". The answer is, groups of growing power, e.g. international software and Internet companies on the one hand, and international pharmaceutical companies on the other - the reason being that these companies, particularly, are interested in a global, stable, level playing field. The object of the present work is to demonstrate theoretically, given the necessary Volterra pair of differential equations, how both the frequency and the amplitude of the individual components could be controlled. As an illustration, the control of frequency was discussed in some detail. The previous publications (Warburton, 1995, 1996) had shown that three financial power blocks could form the basis of such a model. However, a prerequisite for adequate fitness of these proposed block components would be the control of the frequency and amplitude of the financial pools.