AI Magazine Volume 7 Number 1 (1986) ( AAAI) of the crucial terms involved in his analysis, such as "probability" Mauadne 4(4):7-14 falsity of his claims is often impossible to assess. Nute, Donald k. '(1980) Topics in conditional logic Dordrecht, Holland: conceptions upon which his view is based do indeed conform M. Ringle, (Ed.), Philosophical Perspectives in Artificial Intelligence traditional conceptions should not be taken for granted, Terry L. Rankin his observation that "Probability theory is today our primary At hens, Georgia such as "average" and "likely," and therefore it is the most natural language for describing those aspects of (heuristic) performance that we seek to improve" (p. Artificial Intelligence and Robotics: On general theoretical grounds, I think, there are excellent Five Overviews. Busi-reasons to suppose that (a)-(f) are fundamental ness/Technology Books; 1984. Gevarter's work was published by the National that serious difficulties seem to confront the theoretical Bureau of Standards as a set of five volumes, and this book, framework he apparently endorses, where these difficulties published by Business/Technology Books, is Gevarter's are especially severe from an epistemological perspective.

Granting all of this, the only complaint that schema constrains mental ascriptions once a system might be raised is altogether excusable, if not also entirely is specified; but it puts no limit on which systems should minor, i.e., that the material presented might not be so have mental states ascribed to them." "Supertrap" which strikes matches in the presence of gassoaked mice, topples dictionaries on mice, and, of course, Discursively considered, however, and especially for snaps shut whenever mice nibble its bait "These habits the purposes of AI research, these very same strengths can betray a common malevolent thread, which is generalizable be seen as weaknesses from the viewpoint of at least two by (and only by) ascribing a persistent goal: dead mice." Now it clearly aside, ascription is important for AI because it provides was not Pearlis aim to forestall alternative theories or to one more way to detect patterns that might otherwise go justify his own approach in contrast to ...

The Center for Automation and Intelligent Systems monocrystal turbine blades for jet engines that are made Research at Case Western Reserve University, founded by investment casting. Essentially, the part is made by in 1984, provides the setting and the administrative and pouring liquid metal into a ceramic mold, but the environment funding mechanisms for coordinating and focusing the capabilities in which this is done must be tightly controlled. of faculty members and students from many There are several other subprocesses that are also tightly disciplines and departments to deal with significant realworld controlled, such as making the mold. The total process is too complex for a single expert The center serves as an interface between separate system; rather, several different expert systems are needed basic research efforts in the various disciplines and academic and should be coordinated in some way, perhaps by a more departments and the multidisciplinary group efforts global expert system. Currently, we are constructing an needed to deal effectively with nontrivial real problems. Wax patterns appear to be essential for the factory of the future.

Twenty-five years ago I had a dream, a daydream, if you will. A dream shared with many of you. I dreamed of a special kind of computer, which had eyes and ears and arms and legs, in addition to its "brain." I did not dream that this new computer friend would be a means of making money for me or my employer or a help for my country - though I loved my country then and still do, and I have no objection to making money. I did not even dream of such a worthy cause as helping the poor and handicapped of the world using this marvelous new machine. No, my dream was filled with the wild excitement of seeing a machine act like a human being, at least in many ways.

Stuart J. Russell Department of Computer Science Stanford University Stanford, CA 94305 ABSTRACT In the absence of specific relevance information, the traditional assumption in the study of analogy has been that the most similar analogue is most likely to provide the correct solutions; a justification for this assumption has been lacking, as has any relation between the similarity measure used and the probability of correctness of the analogy. We show how a statistical analysis can be performed to give the probability that a given source will provide a successful analogy, using only the assumption that there are some relevant features somewhere in the source and target descriptions. The predicted variation of the probability with source-target similarity corresponds closely to empirical analogy data obtained by Shepard for human and animal subjects for a wide variety of domains. The utility of analogy by similarity seems to rest on some very fundamental assumptions about the nature of our representations.* I INTRODUCTION Analogical reasoning is usually defined as the argument from known similarities between two things to the existence of further similarities.

We show that the notion of bias in inductive concept learning can be quantified in a way that directly relates to learning performance, and that this quantitative theory of bias can provide guidance in the design of effective learning algorithms. We apply this idea by measuring some common language biases, including restriction to conjunctive concepts and conjunctive concepts with internal disjunction, and, uided by these measurements, develop learning algorithms P or these classes of concepts that have provably good convergence properties. Introduction The theme of this pa er is that the notion of bias in inductive concept learning Ip U86] [R86/ can be quantified in a way that enables us to rove meaningful convergence properties for learning algorit i ms. We measure bias with a combinatorial parameter defined on classes of concepts known as the Vapnik-Chervonenkis dimension (or simply d ensiorr) [VC71/, [P78j', JBEHW86/. The lower the dlmenslon of the class of concepts considered by the learning algorithm, the stronger the bias.

The technology for building knowledge-based systems by inductive inference from examples hasbeen demonstrated successfully in several practical applications. This paper summarizes an approach to synthesizing decision trees that has been used in a variety of systems, and it describes one such system, ID3, in detail. Results from recent studies show ways in which the methodology can be modified to deal with information that is noisy and/or incomplete. A reported shortcoming of the basic algorithm is discussed and two means of overcoming it are compared. The paper concludes with illustrations of current research directionsMachine Learning, 1, p. 81-106

The major limitations in building large software have always been (a) its brittleness when confronted by problems that were not foreseen by its builders, and (by the amount of manpower required. The recent history of expert systems, for example highlights how constricting the brittleness and knowledge acquisition bottlenecks are. Moreover, standard software methodology (e.g., working from a detailed "spec") has proven of little use in AI, a field which by definition tackles ill- structured problems. How can these bottlenecks be widened? Attractive, elegant answers have included machine learning, automatic programming, and natural language understanding. But decades of work on such systems have convinced us that each of these approaches has difficulty "scaling up" for want a substantial base of real world knowledge.

This article explores the implications of one type of cognitive technology, techniques and concepts to develop joint human-machine cognitive systems, for the application of computational technology by examining the joint cognitive system implicit in a hypothetical computer consultant that outputs some form of problem solution. This analysis reveals some of the problems can occur in cognitive system design-e.g., machine control of the interaction, the danger of a responsibility-authority double-bind, and the potentially difficult and unsupported task of filtering poor machine solutions. The result is a challenge for applied cognitive psychology to provide models, data, and techniques to help designers build an effective combination between the human and machine elements of a joint cognitive system.

Many of the ideas behind object-oriented programming have roots going back to SIMULA. The first substantial interactive, display-based implementation was the SMALLTALK language. The object-oriented style has often been advocated for simulation programs, systems programming, graphics, and AI programming. The history of ideas has some additional threads including work on message passing as in ACTORS, and multiple inheritance as in FLAVORS. It is also related to a line of work in AI on the theory of frames and their implementation in knowledge representation languages such as KRL, KEE, FRL, and UNITS.