Results


DEVISING HEURISTICS / 23 A PROBLEM SIMILARITY APPROACH TO DEVISING HEURISTICS: FIRST RESULTS

Classics (Collection 2)

John Gaschnig Department of Computer Science Carnegie-Mellon University Pittsburgh, PA 15213 Abstract Here we describe an approach, based upon a notion of problem similarity, that can be used when attempting to devise a heuristic for a given search problem (of a sort represented by graphs). The next step is to find an algorithm for finding paths in P2, then apply this algorithm in a certain way as a heuristic for P1. Using the As algorithm, we experimentally compare the performance of this "maxsort" heuristic for the 8-puzzle with others in the literature. Many combinatorially large problems cannot be solved feasibly by exhaustive case analysis or brute force search, but can be solved efficiently if a heuristic can be devised to guide the search. Research to date on devising heuristics has spanned several problem-solving domains and several approaches.


MACHINE INTELLIGENCE 9

Classics (Collection 2)

Donald Michie Volumes 1 --7 are published by Edinburgh University Press and in the United States by Halsted Press (a subsidiary of John Wiley & Sons, Inc.) Volumes 8 -- 9 are published by Ellis Horwood Ltd., Publishers, Chichester and in the United States by Halsted Press (a subsidiary of John Wiley & Sons, Inc.) MACHINE INTELLIGENCE 9 The publisher's colophon is reproduced from James Gillison's drawing of the ancient Market Cross, Printed in Great Britain by Biddles of Guildford All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted, in any form of by any means, electronic, mechanical, photocopying, recording or otherwise, without prior permission. One intelligent approach to prefaces -- is to have the empty preface. The well prepared reader will form a good idea of the technical programme just from looking at the table of contents; together with the names of the authors, this gives him a good idea of what happened at the symposium. I could try to assess the tallcs and direct the reader's attention to the more interesting communications.


A Bibliography of Computer Chess

Classics (Collection 2)

The following is a fairly comprehensive list of English language articles on computer chess. Although works about related games like checkers and GO have been excluded, it would be wrong not to refer here to A. L. Samuel's early masterpiece "Some studies in machine learning using the game of checkers" In McDonell, whose general assistance was much appreciated. Many people reviewed and commented upon early drafts, the comments and observations by Max Bramer and Hartmut Tanke being especially valuable. Readers may also be interested in the excellent annotated bibliography by Harald Relcsten [259], whose reviews include not only quotations and paraphrased abstracts, but interesting observations. For computer chess works in other languages, especially German and Russian, a revised version of the bibliography by Egbert Meissenberg "Schach liche leistungen von computer", Deutsche Schachblaetter (1968), 1-4, is reputed to be the most correct.


8 A Theory of Advice

Classics (Collection 2)

Machine intelligence problems are sometimes defined as those problems which (i) computers can't yet do, and (ii) humans can. We shall further consider how much "knowledge" about a finite mathematical function can, on certain assumptions, be credited to a computer program. Although our approach is quite general, we are really only interested in programs which evaluate "semihard" functions, believing that the evaluation of such functions constitutes the defining aspiration of machine intelligence work. If a function is less hard than "semihard," then we can evaluate it by pure algorithm (trading space for time) or by pure lookup (making the opposite trade), with no need to talk of knowledge, advice, machine intelligence, or any of those things. We call such problems "standard."


7 Dynamic Probability, Computer Chess, and the Measurement of Knowledge

Classics (Collection 2)

Philosophers and - "pseudognosticians" (the artificial intelligentsial) are coming more and more to recognize that they share common ground and that each can learn from the other. This has been generally recognized for many years as far as symbolic logic is concerned, but less so in relation to the foundations of probability. In this essay I hope to convince the pseudognostician that the philosophy of probability is relevant to his work. Formal systems, such as those used in mathematics, logic, and computer programming, can lead to deductions outside the system only when there is an input of assumptions. For example, no probability can be numerically inferred from the axioms of probability unless some probabilities are assumed without using the axioms: ex nihilo nihil fit.2


1 On Alan Turing and the Origins of Digital Computers

Classics (Collection 2)

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 postwar general purpose computer projects in Britain can be traced back to these wartime computers. During my amateur investigations into computer history, I grew intrigued by the lack of information concerning the role played by the late Alan Turing. His historic paper on computability, and the notion of a universal automaton, had been published in 1936, but there was no obvious connection between these two activities. The purpose of this paper is to document the surprising (and tantalizing) results that I have obtained to date.


BOXES: AN EXPERIMENT IN ADAPTIVE CONTROL

Classics (Collection 2)

BOXES is the name of a computer program. This is what the chess player does when he lumps together large numbers of positions as being'similar' to each other, by neglecting the strategically irrelevant features in which they differ. The resultant small game can be said to be a'model' of the large game. To give a brutally extreme example, consider a specification of chess positions so incomplete as to map from the viewpoint of White the approximately 1050 positions of the large game on to the seven shown in Figure 1. Even this simple classification may have a role in the learning of chess.


NEW DEVELOPMENTS OF THE GRAPH TRAVERSER

Classics (Collection 2)

This paper describes some recent experiments with a computer program which is capable of useful, or at least interesting, application to a number of different problems. The program, the Graph Traverser, has been described in detail in a previous paper (Doran & Michie 1966). However, we shall here need to view the basic algorithm from a rather more general standpoint, corresponding to an actual extension in the flexibility of the program, so that a restatement of what the program can do is desirable. The Graph Traverser, which is written in Elliott 4100 Algol, is potentially applicable to problem situations which can be idealised in the following way (see for comparison Newell and Ernst 1965). There is given a set of'states', which are connected by a set of'transformations', or, as I shall call them, 'operators'.


MACHINE INTELLIGENCE 13

Classics (Collection 2)

OXFORD 1994 Oxford University Press, Walton Street, Oxford 0X2 6DP Oxford New York Athens Auckland Bangkok Bombay Calcutta Cape Town Dar es Salaam Delhi Florence Hong Kong Istanbul Karachi Kuala Lumpur Madras Madrid Melbourne Mexico City Nairobi Paris Singapore Taipei Tokyo Toronto and associated companies in Berlin lbadan Published in the United States by Oxford University Press Inc., New York 0 E. K. Furukawa, D. Michie, and S. Muggleton, 1994 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press.


2 Logic and learning: Turing's legacy

Classics (Collection 2)

Turing's best known work is concerned with whether universal machines can decide the truth value of arbitrary logic formulae. However, in this paper it is shown that there is a direct evolution in Turing's ideas from his earlier investigations of computability to his later interests in machine intelligence and machine learning. Turing realised that machines which could learn would be able to avoid some of the consequences of Godes and his results on incompleteness and undecidability. Machines which learned could continuously add new axioms to their repertoire. Inspired by a radio talk given by Turing in 1951, Christopher Strachey went on to implement the world's first machine learning program.