980 resultados para language evolution
Resumo:
Both earthquake prediction and failure prediction of disordered brittle media are difficult and complicated problems and they might have something in common. In order to search for clues for earthquake prediction, the common features of failure in a simple nonlinear dynamical model resembling disordered brittle media are examined. It is found that the failure manifests evolution-induced catastrophe (EIC), i.e., the abrupt transition from globally stable (GS) accumulation of damage to catastrophic failure. A distinct feature is the significant uncertainty of catastrophe, called sample-specificity. Consequently, it is impossible to make a deterministic prediction macroscopically. This is similar to the question of predictability of earthquakes. However, our model shows that strong stress fluctuations may be an immediate precursor of catastrophic failure statistically. This might provide clues for earthquake forecasting.
Resumo:
A high order difference scheme is used to simulate the spatially developing compressible axisymmetric jet. The results show that the Kelvin-Helmholtz instability appears first when the jet loses its stability, and then with development of jet the increase in nonlinear effects leads to the secondary instability and the formation of the streamwise vortices. The evolution of the three-dimensional coherent structure is presented. The computed results verify that in axisymmetric jet the secondary instability and formation of the streamwise vortices are the important physical mechanism of enhancing the flow mixing and transition occurring.
Resumo:
对LY12铝合金在低周疲劳条件下的裂纹情况和随后进行的动态拉伸条件下裂纹的发展给予了观察和统计分析。发现裂纹的累积数密度分布在损伤演化过程中保持指数形式,用NAG模型对实验结果进行分析,得出该材料裂纹演化发展方程的各种参数。
Discriminative language model adaptation for Mandarin broadcast speech transcription and translation
Resumo:
This paper investigates unsupervised test-time adaptation of language models (LM) using discriminative methods for a Mandarin broadcast speech transcription and translation task. A standard approach to adapt interpolated language models to is to optimize the component weights by minimizing the perplexity on supervision data. This is a widely made approximation for language modeling in automatic speech recognition (ASR) systems. For speech translation tasks, it is unclear whether a strong correlation still exists between perplexity and various forms of error cost functions in recognition and translation stages. The proposed minimum Bayes risk (MBR) based approach provides a flexible framework for unsupervised LM adaptation. It generalizes to a variety of forms of recognition and translation error metrics. LM adaptation is performed at the audio document level using either the character error rate (CER), or translation edit rate (TER) as the cost function. An efficient parameter estimation scheme using the extended Baum-Welch (EBW) algorithm is proposed. Experimental results on a state-of-the-art speech recognition and translation system are presented. The MBR adapted language models gave the best recognition and translation performance and reduced the TER score by up to 0.54% absolute. © 2007 IEEE.
Resumo:
In speech recognition systems language model (LMs) are often constructed by training and combining multiple n-gram models. They can be either used to represent different genres or tasks found in diverse text sources, or capture stochastic properties of different linguistic symbol sequences, for example, syllables and words. Unsupervised LM adaptation may also be used to further improve robustness to varying styles or tasks. When using these techniques, extensive software changes are often required. In this paper an alternative and more general approach based on weighted finite state transducers (WFSTs) is investigated for LM combination and adaptation. As it is entirely based on well-defined WFST operations, minimum change to decoding tools is needed. A wide range of LM combination configurations can be flexibly supported. An efficient on-the-fly WFST decoding algorithm is also proposed. Significant error rate gains of 7.3% relative were obtained on a state-of-the-art broadcast audio recognition task using a history dependently adapted multi-level LM modelling both syllable and word sequences. ©2010 IEEE.
Resumo:
Development of shear bands in saturated soils is a multi-stage process based on the theoretical and numerical investigations in this paper. The soil is initially in homogenous shear strain state, and the instability can be characterized by a dimensionless number D. The inhomogenous distribution of shear strains appears when D>1, and the shear band will initiate and develop gradually. Numerical solutions show that only single shear band that is finally formed in the central region of the specimen even several disturbances (distributed along the specimen) appear in the beginning.
Resumo:
Damage evolution of heterogeneous brittle media involves a wide range of length scales. The coupling between these length scales underlies the mechanism of damage evolution and rupture. However, few of previous numerical algorithms consider the effects of the trans-scale coupling effectively. In this paper, an adaptive mesh refinement FEM algorithm is developed to simulate this trans-scale coupling. The adaptive serendipity element is implemented in this algorithm, and several special discontinuous base functions are created to avoid the incompatible displacement between the elements. Both the benchmark and a typical numerical example under quasi-static loading are given to justify the effectiveness of this model. The numerical results reproduce a series of characteristics of damage and rupture in heterogeneous brittle media.
Resumo:
The thermal stability of nanocrystalline clusters, the phase evolution, and their effects on magnetic Propertieswere studied for as-cast Nd60Al10Fe20Co10 alloy using differential scanning calorimetry curves, x-ray diffraction patterns, scanning electron microscopy, and high-resolution transition electron microscopy. Thermomagnetic curves and hysteresis loops of the bulk metallic glass were measured during the annealing process. The high thermostability of the hardmagnetic properties of the samples observed is attributed to the stability of the nanocrystalline clusters upon annealing, while the slight enhancement in the magnetization is due to the precipitation of some Nd-rich metastable phases. The mechanism of thermostability of the nanocrystalline clusters and the formation of the metastable phases are discussed.
Resumo:
This paper performed a numerical simulation on temperature field evolution for the surface layer of a metallic alloy subjected to pulsed Nd:YAG laser treatment. The enthalpy method was adopted to solve the moving boundary problem, I.e. Stefan problem. Computational results were obtained to show the temperature field evolution. Effects of latent heat and mushy zone width on the temperature field were investigated. The results also show very high values of temperature gradient and cooling rate, which are typical characteristics during the solidification process.
Resumo:
Surface rapid solidification microstructures of AISI 321 austenitic stainless steel and 2024 aluminum alloy have been investigated by electron beam remelting process and optical microscopy observation. It is indicated that the morphologies of the melted layer of both stainless steel and aluminum alloy change dramatically compared to the original materials. Also, the microstructures were greatly refined after the electron beam irradiation.
Resumo:
The process of damage evolution concerns various scales, from micro- to macroscopic. How to characterize the trans-scale nature of the process is on the challenging frontiers of solid mechanics. In this paper, a closed trans-scale formulation of damage evolution based on statistical microdamage mechanics is presented. As a case study, the damage evolution in spallation is analyzed with the formulation. Scaling of the formulation reveals that the following dimensionless numbers: reduced Mach number M, damage number S, stress wave Fourier number P, intrinsic Deborah number D*, and the imposed Deborah number De*, govern the whole process of deformation and damage evolution. The evaluation of P and the estimation of temperature increase show that the energy equation can be ignored as the first approximation in the case of spallation. Hence, apart from the two conventional macroscopic parameters: the reduced Mach number M and damage number S, the damage evolution in spallation is mainly governed by two microdamage-relevant parameters: the Deborah numbers D* and De*. Higher nucleation and growth rates of microdamage accelerate damage evolution, and result in higher damage in the target plate. In addition, the mere variation in nucleation rate does not change the spatial distribution of damage or form localized rupture, while the increase of microdamage growth rate localizes the damage distribution in the target plate, which can be characterized by the imposed Deborah number De*.
Resumo:
Existing devices for communicating information to computers are bulky, slow to use, or unreliable. Dasher is a new interface incorporating language modelling and driven by continuous two-dimensional gestures, e.g. a mouse, touchscreen, or eye-tracker. Tests have shown that this device can be used to enter text at a rate of up to 34 words per minute, compared with typical ten-finger keyboard typing of 40-60 words per minute. Although the interface is slower than a conventional keyboard, it is small and simple, and could be used on personal data assistants and by motion-impaired computer users.
Resumo:
This paper presents models to describe the dislocation dynamics of strain relaxation in an epitaxial uniform layer, epitaxial multilayers and graded composition buffers. A set of new evolution equations for nucleation rate and annihilation rate of threading dislocations is developed. The dislocation interactions are incorporated into the kinetics process by introducing a resistance term, which depends only on plastic strain. Both threading dislocation nucleation and threading dislocation annihilation are characterized. The new evolution equations combined with other evolution equations for the plastic strain rate, the mean velocity and the dislocation density rate of the threading dislocations are tested on GexSi1-x/Si(100) heterostructures, including epitaxial multilayers and graded composition buffers. It is shown that the evolution equations successfully predict a wide range of experimental results of strain relaxation and threading dislocation evolution in the materials system. Meanwhile, the simulation results clearly signify that the threading dislocation annihilation plays a vital role in the reduction of threading dislocation density.