A method for improving classification reliability of multilayerperceptrons

Criteria for evaluating the classification reliability of a neural classifier and for accordingly making a reject option are proposed. Such an option, implemented by means of two rules which can be applied independently of topology, size, and training algorithms of the neural classifier, allows one to improve the classification reliability. It is assumed that a performance function P is defined which, taking into account the requirements of the particular application, evaluates the quality of the classification in terms of recognition, misclassification, and reject rates. Under this assumption the optimal reject threshold value, determining the best trade-off between reject rate and misclassification rate, is the one for which the function P reaches its absolute maximum. No constraints are imposed on the form of P, but the ones necessary in order that P actually measures the quality of the classification process. The reject threshold is evaluated on the basis of some statistical distributions characterizing the behavior of the classifier when operating without reject option; these distributions are computed once the training phase of the net has been completed. The method has been tested with a neural classifier devised for handprinted and multifont printed characters, by using a database of about 300000 samples. Experimental results are discussed.     

Identification of gate electrode discontinuities in submicron CMOStechnologies, and effect on circuit performance

An experiment to determine the effect of gate electrode resistivity on circuit speed gave unexpected results: circuits with the lowest sheet resistance had the poorest circuit speed. Explanation of this behaviour required development of a new high-frequency method of measuring the impedance of the gate electrode. This method revealed that the circuits with a composite gate electrode had been formed with a partial discontinuity. The measurement technique is described, and the evidence of the discontinuity is shown. The effect of the discontinuity on device and circuit speed is demonstrated     

Improving the accuracy of autotransformer and transformer alternating-current bridges

Variable-ratio and computational methods are described for improving the accuracy of autotransformer and transformer ac bridges that take into account and eliminate the additive and multiplicative components of the systematic error due to connecting leads and jumpers of the measuring circuit. Translated from Izmeritel'naya Tekhnika, No. 6, pp. 61–63, June, 1996.     

High-pressure dc sputtering of High-Tc BPSCCO thin films withTc(0) above 100 K

High-T _c Bi(Pb)-Sr-Ca-Cu-O thin films have been made on single-crystal MgO substrates using high-pressure dc sputtering technique. X-ray studies confirm the crystallinity and highly oriented structure withc-axis perpendicular to the substrate. By optimizing the annealing schedule the formation of the high-T _c phase is stabilized. The best film exhibited superconducting transition temperature with zero-resistance temperature,T _c(0), as high as 101 K. Temperature dependence ofJ _c indicates the presence of Josephson-type weak links.     

Self-co-operative ternary pulse-compression sequences

An algorithm called a Hamming scan was developed recently for obtaining sequences with large merit factors and is adopted here to obtain such sequences within which there are nontrivial segments of large merit factors. Correlative detection of the return signal can be based simultaneously on the entire sequence and its segments with large merit factors. Such a coincidence detection scheme can be characterized by a Schur merit factor of the sequence. Sequences with large Schur merit factors are listed.     

Specific heat of a marginal Fermi liquid

The temperature dependence of a marginal Fermi liquid has been calculated. We showed that the expectedT lnT correction is characteristic for the low-temperature domain. The high-temperature domain has a supplementary correction. The results are in agreement with the non-Fermi behavior of some metallic systems the in low-temperature domain.     

Characterization of spherically focused transducers using an ultrasonic measurement model approach

A new and computationally efficient method is developed for characterizing a spherically focused, ultrasonic transducer (and its accompanying test system). Procedures for determining the probe's effective radius, effective focal length, and system efficiency factor are described. Predicted responses that make use of these effective parameters are shown to correspond very well to measured responses for a number of different transducers.     

Surface structure of silver thin films on In2O3:Sn and Al2O 3 †

Surface structure of thin silver films (200 Å) on two technologically important films, indium tin oxide (ITO) and aluminium oxide, has been studied using scanning tunneling microscope. ITO films were prepared by reactive electron beam evaporation. Aluminium oxide films were prepared by oxidizing 2000 Å thick aluminium films evaporated on to H₂ terminated single crystal silicon substrates. The surface structure of silver on ITO and aluminium oxide appeared to be same and was characteristic of Stranski-Krastanov type. The observed asymmetry in the island shape was attributed to the anisotropic nature of the strain fields surrounding the nucleation centres.