The founder of modern computational logic, J.A. Robinson, opens this volume with a chapter on the fields' great forefathers John von Neumann and Alan Turing. Stephen Muggleton follows with an analysis of Turing's legacy in logic and machine learning, conceiving these not in generality, but as specific means of imparting knowledge to computers, a theme first articulated by Turing in the late 1940's. The present volume records the Machine Intelligence Workshop of 1992, held at Strathclyde University's Ross Priory retreat on Loch Lomond, Scotland. Here the series entered not only its second quarter-century but a new phase. As can be seen in these pages, machine learning emerged to declare itself as a seed-bed of new theory, as a practical tool in engineering disciplines, and as material for new metal models in the human sciences. Connections with behavioural and cognitive psychology are illuminated in Chapters 9 and 10. The pioneers always stressed these connections. In 1953 Claude Shannon had this to say: "The problem of how the brain works and how machines may be designed to simulate its activity is surely one of the most important and difficult facing science... Can we organise machines into a hierarchy of levels, as the brain appears to be organised, with the learning of the machine gradually progressing up the hierarchy?... How can computer memory be organised to learn and remember by association, in a manner similar to the human brain?" Approaches to learning by association "in a manner similar to the human brain" have recently engendered unprecedented interest, one might almost say turbulence. Chapter 13 pre-views a joint European endeavour of six academic and six industrial laboratories to steer the topic towards clearer waters. The complete comparative study is now available as a book from Ellis Horwood (Simon and Schuster). January 1994 Stephen Muggleton Executive Editor Donald Michie Koichi Furukawa Associate Editors
New beginnings par excellence also spring from an agreement concluded in 1991 between the Turing Institute, UK and the Japan Society for Artificial Intelligence, Tokyo, under the generous auspices of the Daiwa Anglo-Japanese Foundation. The Foundation provided funding, covering Workshops 13 and 14, to defray travel and attendance costs for six Japanese and six British scientists nominated by the respective parties. To this we owe the circumstance that this volume has been able properly to reflect something of the vigour with which the subject is being advanced in Japan. We are also indebted to the Royal Society of London for facilitating Professor Enn Tyugu's participation from the Estonian Academy of Sciences. Strathclyde University, the Turing Institute, and Scottish Enterprise also contributed help and resource in the many small ways that go towards the making of a great occasion. The Editors would also like to express their thanks to Ashwin Srinivasan for the many hours of effort involved in persuading LaTeX to produce the standard Machine Intelligence look-and-feel within this volume. Thanks are also due to the Oxford University Computing Laboratory for kindly allowing use of printing and document preparation facilities in the production of this volume.
PREFACE by Stephen Muggleton, Donald Michie, and Koichi Furukawa ACKNOWLEDGEMENTS HISTORICAL PERSPECTIVES 1. Logic Computers, Turing, and von Neumann 1 J.A. ROBINSON 2. Logic and learning: Turing's legacy 37 S. MUGGLETON INDUCTIVE INFERENCE 3. A generalization of the least general generalization 59 H. ARIMURA, T. SHINOHARA, S. OTSUKI & H. ISHIZAKA 4. The justification of logical theories based on data compression 87 A. SRINIVASAN, S. MUGGLETON, AND M. BAIN 5. Utilizing structure information in concept formation 123 K. HANDA, M. NISHIKIMI AND H. MATSUBARA 6. The discovery of propositions in noisy data 143 HIROSHI TSUKIMOTO AND CHIE MORITA 7. Learning non-deterministic finite automata from queries 169 and counterexamples T. YOKOMORI SCIENTIFIC DOMAINS 8. Machine Learning and biomolecular modelling 193 M.J.E. STERNBERG, R.A. LEWIS, R.D. KING AND S. MUGGLETON 9. More than meets the eye: animal learning and knowledge induction 213 E.J. KEHOE 10.Regulation of human cognition and its growth 247 C. TREVARTHEN 11.Large heterogeneous knowledge bases 269 E. TYUGU EXPERIMENTAL MACHINE LEARNING 12. Learning optimal chess strategies 291 M. BAIN AND S. MUGGLETON 13.A Comparative study of classification algorithms: Statistical, 311 Machine Learning and Neural Network R.D. KING, R. HENERY, C. FENG AND A. SUTHERLAND LEARNING CONTROL 14. Recent progress with BOXES 363 C. SAMMUT 15. Building symbolic representations of intuitive real-time 385 skills from performance data D. MICHIE AND R. CAMACHO 16. Learning perceptually chunked macro operators 419 M. SUWA AND H. MOTODA 17. Inductively speeding up logic programs 459 M. NUMAO, T. MARUOKA, AND M. SHIMURA
Machine Intelligence 13 - Machine Intelligence and Inductive Learning
Editors:Publisher: Clarendon Press 1994
- K. Furukawa
- Keio University, Tokyo
- Donald Michie
- Turing Institute, Glasgow
- S. Muggleton
- Oxford University Computing Laboratory
Proceedings of the Thirteenth Machine Intelligence Workshop, held at Strathclyde University, 1992.
Up to Machine Intelligence main page.
Go to other volumes: 1 2 3 4 5 6 7 8 9 10 11 12 14