排序方式: 共有37条查询结果,搜索用时 46 毫秒
1.
2.
Sreangsu Acharyya Sumit Negi L. Venkata Subramaniam Shourya Roy 《International Journal on Document Analysis and Recognition》2009,12(3):175-184
Noise in textual data such as those introduced by multilinguality, misspellings, abbreviations, deletions, phonetic spellings,
non-standard transliteration, etc. pose considerable problems for text-mining. Such corruptions are very common in instant
messenger and short message service data and they adversely affect off-the-shelf text mining methods. Most techniques address
this problem by supervised methods by making use of hand labeled corrections. But they require human generated labels and
corrections that are very expensive and time consuming to obtain because of multilinguality and complexity of the corruptions.
While we do not champion unsupervised methods over supervised when quality of results is the singular concern, we demonstrate
that unsupervised methods can provide cost effective results without the need for expensive human intervention that is necessary
to generate a parallel labeled corpora. A generative model based unsupervised technique is presented that maps non-standard
words to their corresponding conventional frequent form. A hidden Markov model (HMM) over a “subsequencized” representation
of words is used, where a word is represented as a bag of weighted subsequences. The approximate maximum likelihood inference
algorithm used is such that the training phase involves clustering over vectors and not the customary and expensive dynamic
programming (Baum–Welch algorithm) over sequences that is necessary for HMMs. A principled transformation of maximum likelihood
based “central clustering” cost function of Baum–Welch into a “pairwise similarity” based clustering is proposed. This transformation
makes it possible to apply “subsequence kernel” based methods that model delete and insert corruptions well. The novelty of
this approach lies in that the expensive (Baum–Welch) iterations required for HMM, can be avoided through an approximation
of the loglikelihood function and by establishing a connection between the loglikelihood and a pairwise distance. Anecdotal
evidence of efficacy is provided on public and proprietary data. 相似文献
3.
Atindra Mohan Bandyopadhyay Aritra Acharyya J. P. Banerjee 《Journal of Computational Electronics》2014,13(3):769-777
A large-signal method based on non-sinusoidal voltage excitation model is used to study the DC and RF characteristics of Double Avalanche Region (DAR) Silicon Transit Time diode. A large-signal simulation program based on drift-diffusion model is developed for this study. The simulation results show the existence of several distinct negative conductance bands in the admittance characteristics separated by positive conductance. Thus the DAR device is capable of delivering RF power not only at the design frequency but also at several frequency bands higher than the design frequency band in the mm-wave regime. A comparative study with DDR Si device designed to deliver RF power at a particular mm-wave frequency band shows that DAR Si device is capable of delivering significantly higher RF power not only at the designed mm-wave frequency band, but also at higher frequency bands. 相似文献
4.
Leak-before-break (LBB) assessment of primary heat transport piping of nuclear reactors involves detailed fracture assessment
of pipes and elbows with postulated throughwall cracks. Fracture assessment requires the calculation of elastic–plastic J-integral and crack opening displacement (COD)1 for these piping components. Analytical estimation schemes to evaluate elastic–plastic J-integral and COD simplify the calculations. These types of estimation schemes are available for pipes with various crack
configurations subjected to different types of loading. However, such schemes for elbow (or pipe bend), which is one of the
important components for LBB analyses, is very meager. Recently, elastic–plastic J and COD estimation scheme has been developed for throughwall circumferentially cracked elbow subjected to closing bending
moment. However, it is well known that the elbow deformation characteristics are distinctly different for closing and opening
bending modes because the ovalisation patterns of elbow cross section are different under these two modes. Development of
elastic–plastic J and COD estimation scheme for an elbow with throughwall circumferential crack at intrados subjected to opening bending moment
forms the objective of the present paper. Experimental validation of proposed J-estimation scheme has been provided by comparing the crack initiation, unstable ductile tearing loads and crack extension
at instability with the test data. The COD estimation scheme has been validated by comparing the COD of test data with the
predictions of the proposed scheme. 相似文献
5.
A generalized analytical model based on multistage scattering phenomena has been developed in this paper for estimating the impact ionization rate of charge carriers in semiconductors. The probabilities of impact ionization initiated by electrons and holes have been calculated separately by taking into account all possible combinations of optical phonon scattering and carrier-carrier collisions prior to the impact ionization. Finally the analytical expressions of impact ionization rate of electrons and holes have been developed by using the aforementioned impact ionization probabilities. The impact ionization rates of electrons and holes in 4H-SiC have been calculated within the field range of \(2.5\times 10^{8}\) – \(6.5\times 10^{8}\hbox { V m}^{-1}\) by using the analytical expressions of those developed in the present paper. Those are also calculated by using the analytical expressions developed by some other researchers earlier without considering the multistage scattering phenomena. Finally the theoretical results obtained from the analytical model proposed in this paper and the analytical model developed by earlier researchers within the field range under consideration have been compared with the ionization rate values calculated by using the empirical relations fitted from the experimentally measured data. Closer agreement with the experimental data has been achieved when the impact ionization rate of charge carriers in 4H-SiC are calculated from the proposed model as compared to the earlier one. 相似文献
6.
Optical control of large-signal properties of millimeter-wave and sub-millimeter-wave DDR Si IMPATTs
Aritra Acharyya Suranjana Banerjee J. P. Banerjee 《Journal of Computational Electronics》2014,13(2):408-424
The authors have proposed a complete large-signal (L-S) model to investigate the optical modulation of high frequency properties of double-drift region (DDR) impact avalanche transit time (IMPATT) devices operating at different millimeter-wave and sub-millimeter-wave frequencies. Simulation is carried out based on the proposed model to study the effect of photo-irradiation of different values of incident optical power of different wavelengths from 600–1000 nm on the DC and L-S characteristics of DDR IMPATTs based on Si designed to operate at 94, 140, 220, 300 and 500 GHz. Two different optical illumination configurations such as top mount and flip chip are taken into account for the present study. Results show that the maximum optical tuning of L-S parameters of the device can be achieved for optical illumination of wavelength 700 nm, i.e. near the wavelength corresponding to the responsivity peak of Si in both top mount and flip chip configurations. Further, better photo-sensitivity of top mount structure is observed as compared to its flip chip counterpart. The simulation results are found to be in good agreement with the experimental results reported earlier. Suitable tunable light source information is also suggested to carry out the optical control experiment within the wavelength range under consideration. 相似文献
7.
A Resistive Random Access Memory (RRAM), where the memory performance principally originated from ‘resistive’ change rather than ‘capacitive’ one (the case with conventional CMOS memory devices), has attracted researchers across the globe, owing to its unique features and advantages meeting the demands of future generation high-speed, ultra low power, nano dimensional memory devices. A large family of semiconducting oxides have been investigated as insulator for Resistive Random Access Memory (RRAM), amongst which TiO2 is one of the potential candidate, principally owing to some of its remarkable advantages e.g. wide band gap, high temperature stability and high dielectric constant with flexibility to offer both unipolar and bipolar switching, which are essential for RRAM device applications. In this review article, we tried to represent the long voyage of TiO2 based RRAM, towards the improvement of the reliability aspects of the device performance in a comprehensive manner. Starting with the key factors like oxygen vacancies, Ti interstitials and electroforming, which are responsible for resistive switching phenomenon, various material preparation techniques for RRAM development have been discussed with emphasis on relative merits and bottlenecks of the process. The factors like electrode material and geometry, device structuring, doping, compliance current, annealing effect etc., which play the pivotal role in determining the switching performance of the device, have been reviewed critically. Finally, the article concludes with the comparison of different TiO2 based RRAM devices followed by the prediction of possible future research trends. 相似文献
8.
9.
10.
This study deals with simulation of low-cycle fatigue (LCF), followed by evaluation of fatigue parameters, which would be
suitable for estimating fatigue lives under uniaxial loading. The cyclic elastic–plastic stress–strain responses were analyzed
using the incremental plasticity procedures. Finite-element (FE) simulation in elastic–plastic regime was carried out in FE
package ABAQUS. Emphasis has been laid on calibration of SS 316 stainless steel for LCF behavior. For experimental verifications,
a series of low-cycle fatigue tests were conducted using smooth, cylindrical specimens under strain-controlled, fully reversed
condition in INSTRON UTM (Universal Testing Machine) with 8,800 controller at room temperature. The comparisons between numerical
simulations and experimental observations reveal the matching to be satisfactory in engineering sense. Based on the cyclic
elastic–plastic stress–strain response, both from experiments and simulation, loop areas, computed for various strain amplitude,
have been identified as fatigue damage parameter. Fatigue strain life curves are generated for fatigue life prediction using
Coffin–Manson relation, Smith–Watson–Topper model, and plastic energy dissipated per cycle (loop area). Life prediction for
LCF has been found out to be almost identical for all these three criteria and correlations between predicted and experimental
results are shown. It is concluded that the improvement of fatigue life prediction depends not only on the fatigue damage
models, but also on the accurate evaluations of the cyclic elastic–plastic stress/strain responses. 相似文献