The application of semantic classification trees to naturallanguage understanding

This article describes a new method for building a natural language understanding (NLU) system, in which the system's rules are learnt automatically from training data. The method has been applied to design of a speech understanding (SU) system. Designers of such systems rely increasingly on robust matchers to perform the task of extracting meaning from one or several word sequence hypotheses generated by a speech recognizer. We describe a new data structure, the semantic classification tree (SCT), that learns semantic rules from training data and can be a building block for robust matchers for NLU tasks. By reducing the need for handcoding and debugging a large number of rules, this approach facilitates rapid construction of an NLU system. In the case of an SU system, the rules learned by an SCT are highly resistant to errors by the speaker or by the speech recognizer because they depend on a small number of words in each utterance. Our work shows that semantic rules can be learned automatically from training data, yielding successful NLU for a realistic application     

A statistical analysis of the received signal from blood duringlaminar flow

In order to determine the limiting velocity resolution that can be achieved using ultrasound, and to provide a model which can be generalized for the analysis of disturbed flow, a theoretical and experimental evaluation of the statistics of the received signal from laminar flow following the transmission of a train of short pulses is presented. The authors derive the autocorrelation function and determine the length of the correlated signal for various flow rates, comparing experimental measurements to theoretical predictions. High resolution experimental RF M-mode images are used to verify the theoretical model. Using a fluid with a density, viscosity, volume concentration, particle size, and speed of sound which is similar to that of blood, the authors show that the signal remains correlated for a long interval under many conditions of clinical interest. Including a comparison with experimental data, the effect of the lateral transit time through the sample volume and the axial velocity spread within the sample volume on the correlation of the received signal is evaluated. When a significant range of velocity components is present within the sample volume, this range is the limiting factor in the length of the correlated signal interval. Therefore, the use of a wideband signal, which reduces the sample volume size, produces a returned signal that may be correlated for a larger number of pulses, or for a longer time     

Nature of terpenoids of aTaxodioxylon gypsaceum from the fossil forest of Dunarobba (Italy)

A specimen fromt he fossil forest of Dunarobba (Central Italy) attributed toTaxodioxylon gypsaceum was chemically analyzed particularly for its terpene content. Main chemical constituents show variations related to ageing and carbonization process, while terpene content exhibits a drop of mono- and sesquiterpenes in favour of diterpenes and higher MW compounds. All components are exclusively hydrocarbons and represent alicyclic or partially aromatic structures. The identified families include abietane, phyllocladane, norabietane and norpimarane among diterpenes and traces of cadinane and selinane among sesquiterpenes.     

XIV. Yeast sequencing reports. Sequence of MKT1, needed for propagation of M2 satellite dsRNA of the L-A virus of Saccharomyces cerevisiae

MKT1 is required for m aintenance of K₂ above 30°C in strains with the L-A-HN variant of the L-A double-stranded RNA virus of Saccharomyces cerevisiae. We report that MKT1 encodes a 92 979 Da protein with serine-rich regions and the retroviral protease signature, DTG, but with no substantial homology to proteins presently in the databases. This sequence is available from GenBank under Accession Number U09129.     

The impact of III-V semiconductors on microelectromechanicalsystems

This article examines the special features of the III-V semiconductors, such as gallium arsenide and chemically related compounds, which make them an attractive alternative to silicon-based technology in the field of microelectromechanical systems. As in conventional electronic systems, it is not expected that the III-V compounds will become exclusively used in all instances, but rather a substantial niche market-exploiting the special advantages of the compound semiconductors-is likely to be established     

Effect of crack-growth mechanism on the prediction of fracture load of adhesive joints

This paper presents observations regarding the cracking behavior of tensile-loaded structural adhesive joints. Experiments showed that fracture occurred by the development and propagation of a damage zone, rather than a single, sharp crack, and that the presence of the adhesive spew fillet did not affect the fracture load of the adhesive joints studied. For joints bonded with the mineral-filled epoxy Cybond 4523GB (American Cyanamid), there was approximately 5 mm of subcritical crack propagation prior to final fracture. Fracture-load predictions based on the initial uncracked geometry made in previous papers were unaffected by this small change in geometry. For joints bonded with the rubber-toughened epoxy Permabond ESP 310, approximately 50 mm of subcritical crack propagation was observed. It was again found that predictions made in previous papers on the basis of the initial geometry gave a good estimate of the final fracture load even though this subcritical crack propagation significantly altered the geometry, and thus the applied energy release rates. The effect of shear deformations of the adherends was also investigated, and it was found that shear deformations could be neglected in engineering calculations for joints subject to remote tensile loading.     

Exact Power Function of Compendial Test Requirements for Content Uniformity

The exact distributions associated with the current compendial test requirements are generated by resorting to the well known Computer Intensive Algorithm method to establish the exact percentage point (limit) for RSD, corresponding to each selected cut-off probability level (confidence level) for each of the four possible experimental outcomes based on the USP-NF test requirements. A table is constructed to present the two-dimensional power function. The similarities between these tabular values and the current compendial RSD limits for 10 and 30 dosage units are extremely remarkable.

Minor differences exist, however. It is suggested that both the theoretical as well as the numerical approaches should be carried out to arrive at a comprehensive solution.     

A multi-purpose CMOS sensor interface for low-power applications

A dedicated low-power CMOS transponder microchip is presented as part of a novel telemetry implant for biomedical applications. This mixed analog-digital circuit contains an identification code and collects information on physiological parameters, i.e., body temperature and physical activity, and on the status of the battery. To minimize the amount of data to be transmitted, a dedicated signal processing algorithm is embedded within its circuitry. All telemetry functions (encoding, modulation, generation of the carrier) are implemented on the integrated circuit. Emphasis is on a high degree of flexibility towards sensor inputs and internal data management, extreme miniaturization, and low-power consumption to allow a long implantation lifetime     

Shunting neural network photodetector arrays in analog CMOS

This paper describes a custom analog CMOS photodetector array IC that exploits nonlinear lateral inhibition to achieve dynamic range compression, edge enhancement, and adaptation to mean input intensity. The neural net array architecture, characterized by nearest-neighbor connections and multiplicative cell interaction, is modeled after biological vision systems. The fabricated IC successfully implements a portion of the compact and powerful nonlinear signal processing performed in the outer layers of the vertebrate retina. Measured results are presented for an optical input intensity range of nearly six decades. A scanning architecture that allows for preferential directional sensitivity is also demonstrated. Measured data agree well with models created using a spreadsheet program