Search results

The purpose of this paper is to report upon research into developing a biologically inspired target‐tracking system (TTS) capable of acquiring quality images of a known target type for a robotic inspection application.

The approach used in the design of the TTS hearkens back to the work on adaptive learning by Oliver Selfridge and Chris J.C.H. Watkins and the work on the classification of objects by Zdzislaw Pawlak during the 1980s in an approximation space‐based form of feedback during learning. Also, during the 1980s, it was Ewa Orlowska who called attention to the importance of approximation spaces as a formal counterpart of perception. This insight by Orlowska has been important in working toward a new form of adaptive learning useful in controlling the behaviour of machines to accomplish system goals. The adaptive learning algorithms presented in this paper are strictly temporal difference methods, including Q‐learning, sarsa, and the actor‐critic method. Learning itself is considered episodic. During each episode, the equivalent of a Tinbergen‐like ethogram is constructed. Such an ethogram provides a basis for the construction of an approximation space at the end of each episode. The combination of episodic ethograms and approximation spaces provides an extremely effective means of feedback useful in guiding learning during the lifetime of a robotic system such as the TTS reported in this paper.

It was discovered that even though the adaptive learning methods were computationally more expensive than the classical algorithm implementations, they proved to be more effective in a number of cases, especially in noisy environments.

The novelty associated with this work is the introduction of an approach to adaptive adaptive learning carried out within the framework of ethology‐based approximation spaces to provide performance feedback during the learning process.

The purpose of this paper is to list and discuss several workers on cybernetics and systems, all fairly recently deceased. Online sources of further details are quoted.

The aim is to review developments on the internet, especially those of general cybernetic interest.

The demise of these workers is a great loss.

For all of the people listed, it is instructive to contemplate their achievements and to speculate on what else they would have done had they lived.

It is hoped this is a valuable periodic review.

The traditional approach to AI is limited because it fails to exploit continuity. The reliance on discrete logic has allowed the rapid initial advance of the subject, but constitutes an inherent deficiency. The limitations have become apparent, and are generally acknowledged by a revival of interest in neural‐net, or connectionist, techniques. This approach has become feasible because of technical developments allowing large‐scale parallel operation. Lessons can be learned by considering the evolution of natural intelligence. Recent studies from a biological viewpoint suggest that this has some unexpected features. The idea of concept formation should be extended to include quantifiable concepts, similar to the semantic variables of fuzzy set theory.

Attention is drawn to a principle of “significance feedback” in neural nets that was devised in the encouraging ambience of the Biological Computer Laboratory and is arguably fundamental to much of the subsequent practical application of artificial neural nets.

The background against which the innovation was made is reviewed, as well as subsequent developments. It is emphasised that Heinz von Foerster and BCL made important contributions prior to their focus on second‐order cybernetics.

The version of “significance feedback” denoted by “backpropagation of error” has found numerous applications, but in a restricted field, and the relevance to biology is uncertain.

Ways in which the principle might be extended are discussed, including attention to structural changes in networks, and extension of the field of application to include conceptual processing.

The original work was 40 years ago, but indications are given of questions that are still unanswered and avenues yet to be explored, some of them indicated by reference to intelligence as “fractal”.

Recalls that the first proposal for artificial neural nets was more than half a century ago. Subsequent developments are reviewed, with particular attention to the interface between regulatory, continuous processing and symbol manipulation. Recalls the standpoint of McCulloch and Pitts, that artificial nets are not precise models but are potentially informative about living systems.

Proposes that halfway through the decade which has been termed that of the brain, our level of understanding is still primitive despite much excellent research. Discusses new findings, such as those presented in a recent lecture, which can still alter profoundly the perception of neural mechanisms, and shows that we may even be wrong in the customary assumption that the well‐known electro‐chemical neural transmission is the only important form of rapid internal communication in the brain.

Building 20 of MIT was erected hurriedly during World War II to house the Radiation Laboratory which was the main US centre for research on radar. Later it housed the Research Laboratory of Electronics of MIT and was the site of a vast amount of innovative research, including much that laid foundations for cybernetics. The unpretentious building was demolished in recent years, and a bizarre and entirely different structure has replaced it. The new design is meant to encourage innovation by a quite different route. The purpose here is to show the importance in the history of cybernetics of what went on in Building 20, which can hardly be overstated, and to argue that for the new building it has to be a “hard act to follow”.

Prompted by an item in the Boston Globe, the old and new buildings are contrasted, with the part played by the old building illustrated by reminiscences.

The reference to the old building as a “magical incubator” is fully warranted.

An instructive contrast is offered between a highly successful but largely fortuitous research environment, and one that is planned in detail and has yet to prove its worth. One valuable planned feature of the old environment, namely a comprehensive “document room”, is described.

The account should be valuable as a historical record.