A boosting algorithm converts a learning machine with error rate less than 50% to one with an arbitrarily low error rate. However, the algorithm discussed here depends on having a large supply of independent training samples. We show how to circumvent this problem and generate an ensemble of learning machines whose performance in optical character recognition problems is dramatically improved over that of a single network. We report the effect of boosting on four databases (all handwritten) consisting of 12,000 digits from segmented ZIP codes from the United State Postal Service (USPS) and the following from the National Institute of Standards and Testing (NIST): 220,000 digits, 45,000 upper case alphas, and 45,000 lower case alphas. We use two performance measures: the raw error rate (no rejects) and the reject rate required to achieve a 1% error rate on the patterns not rejected.

Memory-based classification algorithms such as radial basis functions orK-nearest neighbors typically rely on simple distances (Euclidean, dotproduct ...), which are not particularly meaningful on pattern vectors. More complex, better suited distance measures are often expensive and rather ad-hoc (elastic matching, deformable templates). We propose a new distance measure which (a) can be made locally invariant to any set of transformations of the input and (b) can be computed efficiently. We tested the method on large handwritten character databases provided by the Post Office and the NIST. Using invariances with respect to translation, rotation, scaling,shearing and line thickness, the method consistently outperformed all other systems tested on the same databases.

The methodological underpinnings of AI are slowly changing. Benchmarks, test beds, and controlled experimentation are becoming more common. Although we are optimistic that this change can solidify the science of AI, we also recognize a set of difficult issues concerning the appropriate use of this methodology. We discuss these issues as they relate to research on agent design. We survey existing test beds for agents and argue for appropriate caution in their use. We end with a debate on the proper role of experimental methodology in the design and validation of planning agents.

Remember, these are only the winners. It is reducing customers' software (KBS) vendor were touted as a natural fit for AI I think It is interesting to note that other $200,000 in personnel costs; other not. I believe it is more a sign of the AI techniques, beyond traditional benefits include increased product (downsizing) times and the need for representation and reasoning, are sales from higher customer satisfaction increased visibility for the conference. In I saw many good signs at the conference systems. In particular are multiple addition, AT&T reports increases in that applied AI is alive and uses of fuzzy logic, case-based reasoning, the quality of work produced and job healthy.

Parole board date order received three different parole dates. On the changes, probation judgments, and new laws earliest of these parole dates, he would be eligible and sentencing guidelines enacted each year for release from prison to serve the remainder by the state legislature also affect sentence calculations. of his sentence in the community. Finally, Because offenders are often sentenced because of overcrowding in the prison, Doe under multiple laws, these changes can received a safety valve date, which is a fraction create a complex equation for judges and of his time to serve until parole.

Nestor Inc. (Providence, R.I.) and Intel Corp. (Santa Clara, Cal.) have The US Army Research Lab and the Knowledge Engineering Group of the US delivered the first samples of a Army Ordnance Center and School (Aberdeen Proving Grounds, Md.) have jointly developed, second-generation developed a visual expert system for diagnostics of the Ml tank's turbine engine. A visualization of the East Quayside area, including landscaping, American Medical Laboratories the road network, buildings, and the Tyne Bridge landmark, is being created (Chantilly, Va.) has implemented as a virtual world. Prospective tenants and purchasers will be able to three speech recognition systems to experience a "walk through" of the buildings. Togai InfraLogic (Irvine, Cal.) has been awarded a Phase II Small Business Innovation The three VoicePath systems, developed Research (SBIR) grant by NASA Johnson Space Center to study fuzzy by Kurzweil AI (Waltham, logic control for improving performance of thermal control systems, including Mass.), contain a 50,000-word dictionary, industrial applications such as air conditioning and energy control. Their research is aimed at helping manufacturers Sciaky (Chicago, Ill.), a developer improve their products while trimming production and retooling costs.

It is Lynn Andrea Stein, Carol AAAI's new Secretary/Treasurer.

This article contains an overview of recent and ongoing projects at Boeing's Huntsville Advanced Computing Group (ACG). In addition, it contains an overview of some of the work being conducted by Boeing's Advanced Civil Space Systems Group. One aspect of ACG's charter is to support the efforts of other groups at Boeing. Thus, AI is not considered a stand-alone field but, instead, is considered an area that can be used to find both long- and short-term solutions for Boeing and its customers. All the projects listed here represent a team effort on the part of both ACG researchers and members of other Boeing organizations.

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.

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