Language model cross adaptation for LVCSR system combination

State-of-the-art large vocabulary continuous speech recognition (LVCSR) systems often combine outputs from multiple subsystems developed at different sites. Cross system adaptation can be used as an alternative to direct hypothesis level combination schemes such as ROVER. In normal cross adaptation it is assumed that useful diversity among systems exists only at acoustic level. However, complimentary features among complex LVCSR systems also manifest themselves in other layers of modelling hierarchy, e.g., subword and word level. It is thus interesting to also cross adapt language models (LM) to capture them. In this paper cross adaptation of multi-level LMs modelling both syllable and word sequences was investigated to improve LVCSR system combination. Significant error rate gains up to 6.7% rel. were obtained over ROVER and acoustic model only cross adaptation when combining 13 Chinese LVCSR subsystems used in the 2010 DARPA GALE evaluation. © 2010 ISCA.

Symmetrical 2-D Hermite-Gaussian square launch for high bit rate transmission in multimode fiber links

A 2-D Hermite-Gaussian square launch is demonstrated to show improved systems capacity over multimode fiber links. It shows a bandwidth improvement over both center and offset launches and exhibits ±5 μm misalignment tolerance. © 2011 Optical Society of America.

Dynamic recrystallization induced by plastic deformation at high strain rate in a Monel alloy

An investigation has been made into the plastic deformation behavior of a Monel alloy deformed at high strain rate of 10(5) s(-1) by split Hopkinson bar. The results reveal that there are some equiaxed grains with an average size of 150 nm in diameter in the center of the shear bands, suggesting that this microstructure characteristics be developed by dynamic recrystallization, arising from the deformation and the rapid temperature rise in the band. Analysis shows that the plastic strain rate and the mobile dislocation density play a key role in the new crystallized grain formation and growth. Based on grain boundary energy change and diffusion mechanism, the grain growth kinetics is developed for plastic deformation at a high strain rate.

Mechanical Properties And Constitutive Relationships Of 30Crmnsia Steel Heated At High Rate

In this paper, the mechanical behavior of 30CrMnSiA steel after heating at a high rate are investigated experimentally and theoretically, including a detailed discussion of the effects of strain rate and temperature. Two constitutive models are presented to describe the mechanical response of this material after heating at a high rate, and verified by experimental results. (C) 2007 Elsevier B.V. All rights reserved.

Pattern Recognition With Weighted Complex Networks

In this paper we introduce a weighted complex networks model to investigate and recognize structures of patterns. The regular treating in pattern recognition models is to describe each pattern as a high-dimensional vector which however is insufficient to express the structural information. Thus, a number of methods are developed to extract the structural information, such as different feature extraction algorithms used in pre-processing steps, or the local receptive fields in convolutional networks. In our model, each pattern is attributed to a weighted complex network, whose topology represents the structure of that pattern. Based upon the training samples, we get several prototypal complex networks which could stand for the general structural characteristics of patterns in different categories. We use these prototypal networks to recognize the unknown patterns. It is an attempt to use complex networks in pattern recognition, and our result shows the potential for real-world pattern recognition. A spatial parameter is introduced to get the optimal recognition accuracy, and it remains constant insensitive to the amount of training samples. We have discussed the interesting properties of the prototypal networks. An approximate linear relation is found between the strength and color of vertexes, in which we could compare the structural difference between each category. We have visualized these prototypal networks to show that their topology indeed represents the common characteristics of patterns. We have also shown that the asymmetric strength distribution in these prototypal networks brings high robustness for recognition. Our study may cast a light on understanding the mechanism of the biologic neuronal systems in object recognition as well.

Atomistic Origin Of Rate-Dependent Serrated Plastic Flow In Metallic Glasses

Nanoindentation simulations on a binary metallic glass were performed under various strain rates by using molecular dynamics. The rate-dependent serrated plastic flow was clearly observed, and the spatiotemporal behavior of its underlying irreversible atomic rearrangement was probed. Our findings clearly validate that the serration is a temporally inhomogeneous characteristic of such rearrangements and not directly dependent on the resultant shear-banding spatiality. The unique spatiotemporal distribution of shear banding during nanoindentation is highlighted in terms of the potential energy landscape (PEL) theory.

Molecular Dynamics Study Of The Mechanical Behavior Of Nickel Nanowire: Strain Rate Effects

We present the analysis of uniaxial deformation of nickel nanowires using molecular dynamics simulations, and address the strain rate effects on mechanical responses and deformation behavior. The applied strain rate is ranging from 1 x 10(8) s(-1) to 1.4 x 10(11) s(-1). The results show that two critical strain rates, i.e., 5 x 10(9) s(-1) and 8 x 10(10) s(-1), are observed to play a pivotal role in switching between plastic deformation modes. At strain rate below 5 x 10(9) s(-1), Ni nanowire maintains its crystalline structure with neck occurring at the end of loading, and the plastic deformation is characterized by {111} slippages associated with Shockley partial dislocations and rearrangements of atoms close to necking region. At strain rate above 8x10(10) s(-1), Ni nanowire transforms from a fcc crystal into a completely amorphous state once beyond the yield point, and hereafter it deforms uniformly without obvious necking until the end of simulation. For strain rate between 5 x 10(9) s(-1) and 8 x 10(10) s(-1), only part of the nanowire exhibits amorphous state after yielding while the other part remains crystalline state. Both the {111} slippages in ordered region and homogenous deformation in amorphous region contribute to the plastic deformation. (C) 2007 Published by Elsevier B.V.

A Modified Free Volume Model For Characterizing Of Rate Effect In Bulk Metallic Glasses

We investigate the plastic deformation and constitutive behaviour of bulk metallic glasses (BMGs). A dimensionless Deborah number De(ID) = t(r)/t(i) is proposed to characterize the rate effect in BMGs, where t(r) is the structural relaxing characteristic time of BMGs under shear load, t(i) is the macroscopic imposed characteristic time of applied stress or the characteristic time of macroscopic deformation. The results demonstrate that the modified free volume model can characterize the strain rate effect in BMGs effectively.