INTRODUCTION Further progress in the application of computers to many practical fields seems to depend heavily on the success in implementing learning and inductive processes within machines. For example, to develop a consultation system for medical or plant disease diagnosis, prognosis and decision making in general, it is very desirable, perhaps even necessary, to be able to'teach' the system through examples of correct and/or incorrect decisions, rather than by precisely describing the decision process in its full generality and then transforming this description into a computer program. A similar situation exists in computer chess. The development of computer programs playing at the master level (especially the end games) seems to be a formidable task if the programs are not eventually able to learn and improve on their decision making rules through the specific examples of games, rather than by being explicitly told all the rules. Due to easy access to human knowledge about chess and the relative simplicity of testing the results, chess is one of the most attractive testing domains for inductive inference programs. This report presents first results from an experiment on the application of an inductive learning program called AQVAL/1 developed at the University of Illinois, to chess end games.

I. INTRODUCTION Describing scientific theory formation as an information-processing problem suggests breaking the problem into subproblems and searching solution spaces for plausible items in the theory. A computer program called meta-DEN D RAL embodies this approach to the theory formation problem within a specific area of science. Scientific theories are judged partly on how well they explain the observed data, how general their rules are, and how well they are able to predict new events. The meta-D END RA L program attempts to use these criteria, and more, as guides to formulating acceptable theories. The problem for the program is to discover conditional rules of the form S-421, where the S's are descriptions of situations and the A's are descriptions of actions. The rule is interpreted simply as'When the situation S occurs, action A occurs'. The theory formation program first generates plausible A's for theory sentences, then for each A it generates plausible S's. At the end it must integrate the candidate rules with each other and with existing theory. In this paper we are concerned only with the first two tasks: data interpretation (generating plausible A's) and rule formation (generating plausible S's for each A). This paper describes the space of actions (A's), the space of situations (S's) and the criteria of plausibility for both. This requires mentioning some details of the chemical task since the generators and the plausibility criteria gain their effectiveness from knowledge of the task. The theory formation task As in the past, we prefer to develop our ideas in the context of a specific task area.

B.RANDELL 3 PROGRAM PROOF AND MANIPULATION 2 Some techniques for proving correctness of programs which alter data structures.

Among the many properties ascribed to the magical number seven is that of marking out the significant epochs of a human life-span -- seven years from birth to departure from the kindergarten, another seven to puberty, another seven to majority. The occasion of the Seventh International Machine Intelligence Workshop is perhaps an appropriate moment to take stock. Views differ as to exactly which of the successive thresholds is the one on which machine intelligence research is now poised. But there is no mistaking the sense of transition, felt both by its practitioners, who claim that their field is at last attaining maturity, and in a rather different way by its interested Spectators. The latter rightly point out that if maturation brings opportunity and new powers it also brings the obligation to earn a living.

Oberuc 245, 249 see predicate calculus Occam's razor 271 London 24-5, 52

The study of perception is divided among many established sciences: physiology, experimental psychology and machine intelligence; with several others making contributions. But each of the contributing sciences tends to have its own concepts, and ways of considering problems. Each -- to use T. S. Kuhn's term (1962) -- has its own'paradigm', within which its science is respectable. This can make co-operation difficult, as misunderstandings (and even distrust) can be generated by paradigm differences. This paper is a plea to consider perceptual phenomena from many points of view, and to consider whether a general paradigm for perception might be found.

This paper documents an investigation into the role that the late Alan Turing played in the development of electronic computers. Evidence is presented that during the war he was associated with a group that designed and built a series of special purpose electronic computers, which were in at least a limited sense'program controlled', and that the origins of several post-war general purpose computer projects in Britain can be traced back to these wartime computers. INTRODUCTION During my amateur investigations into computer history, I grew intrigued by the lack of information concerning the role played by the late Alan Turing.

C. COOPER 21 3 Data representation--the key to conceptualisation: D. B. VIGOR 33 MECHANISED MATHEMATICS 45 4 An approach to analytic integration using ordered algebraic expressions: L. I. HODGSON 47 5 Some theorem-proving strategies based on the resolution principle: J. L DARLINGTON 57 MACHINE LEARNING AND HEURISTIC PROGRAMMING 73 6 Automatic description and recognition of board patterns in Go-Moku: A. M. MURRAY and E. W. Etcomc

BOXES is the name of a computer program. We shall also use the word boxes to refer to a particular approach to decision-taking under uncertainty which has been used as the basis of a number of computer programs. Figure 1 shows a photograph of an assemblage of actual boxes--matchboxes to be exact. Although the construction of this Matchbox Educable Noughts and Crosses Engine (Michie 1961, 1963) was undertaken as a'fun project', there was present a more serious intention to demonstrate the principle that it may be easier to learn to play many easy games than one difficult one. Consequently it may be advantageous to decompose a game into a number of mutually independent sub-games even if much relevant information is put out of reach in the process. The principle is related to the method of subgoals in problem-solving (see Newell et al. 1960) but differs in one fundamental respect: subgoals are linked in series, while sub-games are played in parallel, in a sense which will become apparent. DECOMPOSITION INTO SUB-GAMES The motivation for developing algorithms for small games (by a'small' game we mean one with so few board positions that a boxes approach is feasible) needs explanation, since small games are generally too trivial to be of intellectual interest in themselves.

JUSTIFICATION OF CHESS PROGRAMS Young animals play games in order to prepare themselves for the business of serious living, without getting hurt in the training period. Game-playing on computers serves a similar function. It can teach us something about the structure of thought processes and the theory of struggle and has the advantage over economic modelling that the rules and objectives are clear-cut. If the machine wins tournaments it must be a good player. The complexity and originality of a master chess player is perhaps greater than that of a professional economist. The chess player continually pits his wits against other players and the precision of the rules makes feasible a depth of thinking comparable to that in mathematics. No program has yet been written that plays chess of even good amateur standard. A really good chess program would be a breakthrough in work on machine intelligence, and would be a great encouragement to workers in other parts of this field and to those ...