The Ninth International Conference on Machine Learning was held in Aberdeen, Scotland, from 1-3 July 1992, with 198 participants in attendance. The conference covered a broad range of topics drawn from the general area of machine learning, including concept-learning algorithms, clustering, speedup learning, formal analysis of learning systems, neural networks, genetic algorithms, and applications of machine learning. This article briefly touches on six selected talks that were of exceptional interest.

The system has reduced time The system, developed by Stereo-spent by store personnel on the telephone seeking answers to point-of-sale Graphics (San Rafael, Calif.), uses a technical problems, and it allows the help desk analysts to handle a wider range video projector equipped with a of responsibilities for the company. Viewers wear passive eye Swedish stock exchange, has developed an intelligent system to advise on how wear which allows each eye to view to deal in stocks and shares. The company's "hit rate" of dealing correctly with the appropriate image, thereby providing stocks has reportedly increased from 60% to 90%. IntelliCorp Inc. (Mountain View, Calif.), an expert system vendor, and James Inference Corp. (El Segundo, Calif.), Martin & Co. (Reston, Va.), a computer-aided software engineering (CASE) consulting a supplier of expert system development group, have launched a field test program for an object-oriented information tools, has teamed up with IDS engineering environment. The new product, called Object Management Financial Services (Minneapolis, Workbench, directly supports rapid application development, and will either Minn.) to jointly develop Macintosh generate a system immediately or draw information from currently available versions of Inference products.

The first Robotics Exhibition and Competition sponsored by the Association for the Advancement of Artificial Intelligence was held in San Jose, California, on 14-16 July 1992 in conjunction with the Tenth National Conference on AI. This article describes the history behind the competition, the preparations leading to the competition, the threedays during which 12 teams competed in the three events making up the competition, and the prospects for other such competitions in the future.

The field of AI is directed at the fundamental problem of how the mind works; its approach, among other things, is to try to simulate its working -- in bits and pieces. History shows us that mankind has been trying to do this for certainly hundreds of years, but the blooming of current computer technology has sparked an explosion in the research we can now do. The center of AI is the wonderful capacity we call learning, which the field is paying increasing attention to. Learning is difficult and easy, complicated and simple, and most research doesn't look at many aspects of its complexity. However, we in the AI field are starting. Let us now celebrate the efforts of our forebears and rejoice in our own efforts, so that our successors can thrive in their research. This article is the substance, edited and adapted, of the keynote address given at the 1992 annual meeting of the Association for the Advancement of Artificial Intelligence on 14 July in San Jose, California. AI Magazine 14(2): 36-48.

In VISION-21 Symposium. NASA Lewis Research Center and the Ohio Aerospace Institute.

What skill is more important to teach than reading?

Knowledge-based natural language processing systems have achieved good success with certain tasks but they are often criticized because they depend on a domain-specific dictionary that requires a great deal of manual knowledge engineering. This knowledge engineering bottleneck makes knowledge-based NLP systems impractical for real-world applications because they cannot be easily scaled up or ported to new domains. In response to this problem, we developed a system called AutoSlog that automatically builds a domain-specific dictionary of concepts for extracting information from text. Using AutoSlog, we constructed a dictionary for the domain of terrorist event descriptions in only 5 person-hours. We then compared the AutoSlog dictionary with a handcrafted dictionary that was built by two highly skilled graduate students and required approximately 1500 person-hours of effort. We evaluated the two dictionaries using two blind test sets of 100 texts each. Overall, the AutoSlog dictionary achieved 98% of the performance of the handcrafted dictionary. On the first test set, the Auto-Slog dictionary obtained 96.3% of the performance of the handcrafted dictionary. On the second test set, the overall scores were virtually indistinguishable with the AutoSlog dictionary achieving 99.7% of the performance of the handcrafted dictionary.

Computational Linguistics, 19 (2), 313-30.

G/SPLINES is an algorithm for building functional models of data. It uses genetic search to discover combinations of basis functions which are then used to build a least-squares regression model. Because it produces a population of models which evolve over time rather than a single model, it allows analysis not possible with other regression-based approaches. 1 INTRODUCTION G/SPLINES is a hybrid of Friedman's Multivariable Adaptive Regression Splines (MARS) algorithm (Friedman, 1990) with Holland's Genetic Algorithm (Holland, 1975). G/SPLINES has advantages over MARS in that it requires fewer least-squares computations, is easily extendable to non-spline basis functions, may discover models inaccessible to local-variable selection algorithms, and allows significantly larger problems to be considered. These issues are discussed in (Rogers, 1991). This paper begins with a discussion of linear regression models, followed by a description of the G/SPLINES algorithm, and finishes with a series of experiments illustrating its performance, robustness, and analysis capabilities.

An important issue in neural computation is the dynamic range of weights in the neural networks. Many experimental results on learning indicate that the weights in the networks can grow prohibitively large with the size of the inputs. Here we address this issue by studying the tradeoffs between the depth and the size of weights in polynomial-size networks of linear threshold elements (LTEs). We show that there is an efficient way of simulating a network of LTEs with large weights by a network of LTEs with small weights. To prove these results, we use tools from harmonic analysis of Boolean functions.