One of the most common mechanisms used for speeding up problem solvers is macro-learning. Macros are sequences of basic operators acquired during problem solving. Macros are used by the problem solver as if they were basic operators. The major problem that macro-learning presents is the vast number of macros that are available for acquisition. Macros increase the branching factor of the search space and can severely degrade problem-solving efficiency. To make macro learning useful, a program must be selective in acquiring and utilizing macros. This paper describes a general method for selective acquisition of macros. Solvable training problems are generated in increasing order of difficulty. The only macros acquired are those that take the problem solver out of a local minimum to a better state. The utility of the method is demonstrated in several domains, including the domain of NxN sliding-tile puzzles. After learning on small puzzles, the system is able to efficiently solve puzzles of any size.

This article reports on the development and implementation of DISXPERT, an intelligent rule-based system tool for referral of social security disability recipients to vocational rehabilitation services. The growing use of paraprofessionals as caseworkers responsible for assessment in the social services area provides fertile domain areas for new and innovative application of intelligent system technology. The main function of DISXPERT is to provide support to paraprofessional caseworkers in reaching unbiased and consistent assessment decisions regarding referral of clients to vocational rehabilitation services. The results after four years of use demonstrate that paraprofessionals using DISXPERT can make assessments in less time and with a level of accuracy superior to the vocational rehabilitation domain professionals using manual methods. This article discusses the problem domain, the design and development of the system, uses of AI technology, payoffs, and deployment and maintenance of the system.

Deneb Robotics (Auburn Hills, Mich.) has been awarded a $2.3 million contract from the National Institute of Standards and Technology (NIST) to develop the agent network for task scheduling and execution. This intelligent agent-based project is designed to improve existing factory-scheduling systems with a new task scheduling and execution system in which Shell U.K. Exploration and Production availability and prevent cars from agents represent factory resources, systems, (Aberdeen, U.K.) has implemented being damaged while they are parked. The Arvin Industries (Columbus, Ind.) is Cisco Systems (San Jose, Calif.), a supplier expert system helped Shell achieve working with the U.S. Air Force to of network technology, is using over $1.6 million in cost savings for develop a neural network system that intelligent-agent technology to integrate its Brent Field site within 2 months of can determine the quality of noise in CD-ROM and online web information implementation. The neural network will help The addition of intelligent The National Research Council has determine what exactly an annoying search-and-retrieval capabilities has awarded Nestor (Providence, R.I.) a sound is and how it can be fixed. Mercedes-Benz plans This system has helped cut specialty Neural Computer Sciences (NCS) to establish three vrf test sites in clinic costs by 40 percent.

AI theory and its technology is rarely consulted in attempted resolutions of social problems. Solutions often require that decision-analytic techniques be combined with expert systems. The emerging literature on combined systems is directed at domains where the prediction of human behavior is not required. A foundational shift in AI presuppositions to intelligent agents working in collaboration provides an opportunity to explore efforts to improve the performance of social institutions that depend on accurate prediction of human behavior. Professionals concerned with human outcomes make decisions that are intuitive or analytic or some combination of both. The relative efficacy of each decision type is described. Justifications and methodology are presented for combining analytic and intuitive agents in an expert system that supports professional decision making. Psychological grounds for the allocation of functions to agents are reviewed. Jury selection, the prototype domain, is described as a process typical of others that, at their core, require the prediction of human behavior. The domain is used to demonstrate the formal components, steps in construction, and challenges of developing and testing a hybrid system based on the allocation of function. The principle that the research taught us about the allocation of function is "the rational and predictive primacy of a statistical agent to an intuitive agent in construction of a production system." We learned that the reverse of this principle is appropriate for identifying and classifying human responses to questions and generally dealing with unexpected events in a courtroom and elsewhere. This principle and approach should be paradigmatic of the class of collaborative models that capitalizes on the unique strengths of AI knowledge-based systems. The methodology used in the courtroom is described along with the history of the project and implications for the development of related AI systems. Empirical data are reported that portend the possibility of impressive predictive ability in the combined approach relative to other current approaches. Problems encountered and those remaining are discussed, including the limits of empirical research and standards of validation. The system presented demonstrates the challenges and opportunities inherent in developing and using AI-collaborative technology to solve social problems.

The following investigates the use of single-neuron learning algorithms to improve the performance of text-retrieval systems that accept natural-language queries. A retrieval process is explained that transforms the natural-language query into the query syntax of a real retrieval system: the initial query is expanded using statistical and learning techniques and is then used for document ranking and binary classification. The results of experiments suggest that Kivinen and Warmuth's Exponentiated Gradient Descent learning algorithm works significantly better than previous approaches. 1 Introduction The following work explores two learning algorithms - Least Mean Squared (LMS) [1] and Exponentiated Gradient Descent (EG) [2] - in the context of text-based Information Retrieval (IR) systems. The experiments presented in [3] use connectionist to improve the retrieval of relevant documents from a largelearning models collection of text. Previous the area employs various techniques for improving retrieval [6, 7, 14].

We describe the implementation of a hidden Markov model state decoding system, a component for a wordspotting speech recognition system.The key specification for this state decoder design is microwatt power dissipation; this requirement led to a continuoustime, analogcircuit implementation. We characterize the operation of a 10-word (81 state) state decoder test chip.

The following investigates the use of single-neuron learning algorithms to improve the performance of text-retrieval systems that accept natural-language queries. A retrieval process is explained that transforms the natural-language query into the query syntax of a real retrieval system: the initial query is expanded using statistical and learning techniques and is then used for document ranking and binary classification. The results of experiments suggest that Kivinen and Warmuth's Exponentiated Gradient Descent learning algorithm works significantly better than previous approaches. 1 Introduction The following work explores two learning algorithms - Least Mean Squared (LMS) [1] and Exponentiated Gradient Descent (EG) [2] - in the context of text-based Information Retrieval (IR) systems. The experiments presented in [3] use connectionist learning models to improve the retrieval of relevant documents from a large collection of text. Previous work in the area employs various techniques for improving retrieval [6, 7, 14].

We describe the implementation of a hidden Markov model state decoding system, a component for a wordspotting speech recognition system. The key specification for this state decoder design is microwatt power dissipation; this requirement led to a continuoustime, analog circuit implementation. We characterize the operation of a 10-word (81 state) state decoder test chip.

Busey Bank (Champaign, Ill.) is using intelligent-agent technology to launch its Lloyds Bowmaker Motor Finance (Petersfield, U.K.) has implemented a The Philadelphia Stock Exchange care products, has developed a rulebased neural network-based system for credit (Philadelphia, Pa.) has adopted an multinational order-entry and scoring new loan applications. The company is system helps Lloyds determine whether increase the reliability and scalability using the system to process orders to accept a loan and gives the reasons of network-supported options-trading from its network of more than for its choice. The system uses an electronic facilities. The software will permit installed a rule-based expert system to camera to image the front face of letters, team members in different geographic manage the complexity of producing identify the destination address, locations to explore similar multisensory more than 20,000 new designs and and determine its delivery-point bar environments both independently 2.4 billion greeting cards annually. The company has completely reengineered its operation, converting an Telecommunications providers MCI Healthcare software developer HBO & antiquated job-shop operation into a (Washington, D.C.) and BT (London, Company (Atlanta, Ga.) is developing state-of-the-art cellular one.

In recent years, there has been a resurgence in research on empirical methods in natural language processing. These methods employ learning techniques to automatically extract linguistic knowledge from natural language corpora rather than require the system developer to manually encode the requisite knowledge. The current special issue reviews recent research in empirical methods in speech recognition, syntactic parsing, semantic processing, information extraction, and machine translation. This article presents an introduction to the series of specialized articles on these topics and attempts to describe and explain the growing interest in using learning methods to aid the development of natural language processing systems.