A wavelet neural network for the detection of heart valve diseases

An expert system is presented for interpretation of the Doppler signals of heart valve diseases based on pattern recognition. We deal in particular with the combination of feature extraction and classification from measured Doppler signal waveforms at the heart valve using Doppler ultrasound. A wavelet neural network model developed by us is used. The model consists of two layers: a wavelet layer and a multilayer perceptron. The wavelet layer used for adaptive feature extraction in the time–frequency domain is composed of wavelet decomposition and wavelet entropy. The multilayer perceptron used for classification is a feedforward neural network. The performance of the developed system has been evaluated in 215 samples. The test results show that this system is effective to detect Doppler heart sounds. The classification rate averaged 91% correct for 123 test subjects.     

Bidirectional evolutionary heuristic for the minimum vertex-cover problem

In this study, we have solved the minimum vertex-cover problem, NP-hard, by the means of genetic algorithms (GA). In prior studies, initial population on which GA iteratively run, was created randomly. However, we have improved this technique to create initial population and the population created has some useful properties such as decreasing the number of iterations of GA algorithm and gets rid of diversity problem, local solution problem. Initially, one chromosome is created and then the inversion of this chromosome is taken as another chromosome. If it is required to create an initial population with large set of chromosomes, then randomly created chromosomes can be divided into desired partitions and other chromosomes can be obtained by the complement of each partition at a time. Search is handled in bi-directional manner, and the initial population contains both ends of solution space.     

Fully nonlinear oscillator noise analysis: an oscillator with no asymptotic phase

Oscillators exist in many systems. Detailed and correct characterization and comprehension of noise in autonomous systems such as oscillators is of utmost importance. Previous approaches to oscillator noise analysis are based on some kind of perturbation analysis, some linear and some nonlinear. However, the derivations of the equations for perturbation analysis are all based on information that is produced by a linearization of the oscillator equations around the periodic steady‐state solution, where it is assumed that the oscillator is orbitally stable and it has the so‐called asymptotic phase property. In this paper, we first discuss these notions from a qualitative perspective, and demonstrate that the asymptotic phase property is crucial in validating all of the previous approaches. We then present the case of a simple oscillator that is orbitally stable but without asymptotic phase, for which previous approaches fail. We then present a fully nonlinear noise analysis of this oscillator. We derive and compute nonlinear, non‐stationary and non‐Gaussian stochastic characterizations for both amplitude and phase noise. We arrive at results that are distinctly different when compared with the ones obtained previously for oscillators with asymptotic phase. We compare and verify our analytical results against extensive Monte Carlo simulations. Copyright © 2006 John Wiley & Sons, Ltd.     

A comparative study of borided pure niobium, tungsten and chromium

Pure niobium (Nb), tungsten (W) and chromium (Cr) were boronized at 940 °C for 2, 4 and 8 h. The borided samples were characterized by X-ray diffraction, Scanning electron microscope and microhardness tests. Tribological investigation was conducted. X-ray study showed the presence of NbB₂, WB, and CrB. The hardnesses of boride layers formed on the pure Nb, W and Cr were 2500, 2500 and 1700 HV, respectively, whereas the hardnesses of the pure Nb, pure W and pure Cr were 110, 445 and 115 HV, respectively. Nb boride layers ranged in thickness from 8 to 22 μm, whereas W boride layers ranged in thickness from 10 to 42 μm, and the thickness of Cr boride layer varied from 4 to 12 μm with boronizing time. The boriding of W resulted in thicker boride layer compared to the boriding of Nb and Cr at given time. The frictional behaviour and wear mechanicms differ in modes and scales.     

Error Diagnosis in Equivalence Checking of High Performance Microprocessors

We describe techniques for diagnosing errors in formal equivalence checking of RTL and transistor level models of high performance microprocessors at Freescale Semiconductor Inc. We use Symbolic Trajectory based Evalaution (STE) for combinational equivalence checking. STE accurately captures transistor level behaviors. We use simulation based error diagnosis techniques and present a seamless integration of them in our current verification environments.     

Real-time X-ray studies of the growth of Mo-seeded Si nanodots by low-energy ion bombardment

Silicon surface evolution during room temperature low-energy (300, 500 and 1000 eV) normal incidence Ar⁺ ion bombardment in the presence of Mo seed atoms has been studied with real-time grazing-incidence small-angle X-ray scattering and ex situ atomic force microscopy. When a small amount of Mo atoms was supplied to the Si surface during ion bombardment, the development of correlated structures with two different characteristic length scales was observed. The shorter length scale features (“dots”) coarsened with time until they reached a constant spatial wavelength. The longer length scale corrugations associated with kinetic roughening, however, continued to grow in amplitude during bombardment. The overall roughness is dominated by different corrugations at different times in the kinetic evolution, showing a complex behavior. The evolution of the kinetic roughening can be described by the Family-Vicsek scaling hypothesis, but measured scaling exponents are not in agreement with those of existing models.