Current moIecuIar bioIogy techniques were deveIoped primarily for characterization of single genes, not entire genomes, and, as such, are not ideally suited to high resolution analysis of complex traits and the moiecular genetics of very large populations.Despite rapid progress in the human genome project effort, there is little doubt that radicaIIy new conceptual approaches are needed before routine whole genome-based analyses can be undertaken by both basic research and clinical laboratories.Physical mapping of genomes, using restriction endonucleases, has played a major role in the identification and characterizing various loci, for example, by aiding clone contig formation and by characterizing genetic lesions.Restriction maps provide precise genomic distances, unlike ordered sequencebased landmarks such as Sequence Tagged Sites (ST%), that are essential for optimizing the efficiency of sequencing efforts, and for determining the spatial relationships of specific loci.When compared to tedious hybridization-based fingerprinting approaches, ordered restriction maps offer relatively unambiguous clone characterization that is useful in contig formation, establishment of minimal tiling paths for sequencing, and preliminary characterization of sequence lesions.In addition, such maps provide a useful scaffold for sequence assembly, often critical in the final sequence finishing stage.Despite the broad applications of restriction maps, the associated techniques for their generation have changed little over the last ten years, primarily because they still utilize electrophoretic analysis.To help overcome these shortcomings, our laboratory developed the first practical non-electrophoretic genomic mapping approach, Optical Mapping, to meet this need.Optical Mapping is a single molecule methodology for the rapid production of ordered restriction

There are many large system problems that are hard to model exactly or in a computationally tractable fashion. Examples include the mapping of human DNA, speech recognition, and automated learning in computer chess. Traditional artificial intelligence solution techniques for such problems rely on a combination of custom encoding of expert knowledge and heuristic search. They take much time to hand craft and then often are unable to take advantage of faster computers as they become available. In this context, the authors explore the advantage of using statistical search techniques in which the knowledge is encoded in some form of statistical model whose parameters are automatically adjusted or trained with domain data. The benefits are faster development times, greater solution accuracy (compared to hand crafted solutions) and the ability to allow the problem size and desired solution accuracy to be scaled up with computational resources. They apply this approach to certain critical computational problems in mapping the human genome. They use a Bayesian model to provide the best solution accuracy as a function of the number of parameters. Heuristic search techniques derived from artificial intelligence are used to search the model space in an efficient manner in the average case.

BEAM is a hardware accelerator that has been designed and built for real-time execution of the SPHINX speaker-independent, continuous-speech recognition system and similar systems. SPHINX on BEAM is able to recognize sentences from a 1000-word vocabulary and a perplexity-60 grammar in about 1.3 times real time. BEAM does not use any custom integrated circuits. The architecture of the accelerator is described. Performance data are given and compared with those for other architectures. It is concluded that BEAM demonstrates how general-purpose technology can be used to build systems that are substantially faster than general-purpose systems.>

This paper describes two custom architectures tailored to a speech recognition beam search algorithm. Both architectures have been simulated using real data and the results of the simulation are presented. The paper also describes the design process of the custom architectures and presents a number of ideas on the automatic design of custom systems for data dependent computations.

The goal of the paper is to present some of the design characteristics and performance of a special purpose custom machine that has the potential of improving the speed of the implementation of a beam search algorithm by two orders of magnitude when compared with a general purpose architecture implementation. The paper also describes the architecture of a general purpose self-timed device that can be used in implementing such a machine.

Path expressions were originally proposed by Campbell and Habermann [1] as a mechanism for process synchronization at the monitor level in software. Not unexpectedly, they also provide a useful notation for specifying the behavior of asynchronous circuits. Motivated by this potential application we investigate how to directly translate path expressions into hardware.

The design of a custom VLSI circuit that has the potential of achieving a high throughput when executing one of the most computationally expensive modules of a speech recognition system is presented. The architecture of the device is described in terms, of a set of self-timed building blocks that implement arithmetic operations, storage and input/output functions. The formalism used to describe the architecture makes it easy to define variations of the basic structure in order to deal with recognition systems that use different heuristics and search different kinds of databases. A tool that helps the designer investigate the area/speed trade-offs is also described.