arXiv subjects

Dong Wang

Publications and source records attributed to Dong Wang.

At least 757 records · Page 42Linked to original sources

arXiv2016

Modular knowledge systems accelerate human migration in asymmetric random environments

Migration is a key mechanism for expansion of communities. In spatially heterogeneous environments, rapidly gaining knowledge about the local environment is key to the evolutionary success of a migrating population. For historical human migration, environmental heterogeneity was naturally asymmetric in the north-south (NS) and east-west (EW) directions. We here consider the human migration process in the Americas, modeled as random, asymmetric, modularly correlated environments. Knowledge about the environments determines the fitness of each individual. We present a phase diagram for asymmetry of migration as a function of carrying capacity and fitness threshold. We find that the speed of migration is proportional to the inverse complement of the spatial environmental gradient, and in particular we find that north-south migration rates are lower than east-west migration rates when the environmental gradient is higher in the north-south direction. Communication of knowledge between individuals can help to spread beneficial knowledge within the population. The speed of migration increases when communication transmits pieces of knowledge that contribute in a modular way to the fitness of individuals. The results for the dependence of migration rate on asymmetry and modularity are consistent with existing archaeological observations. The results for asymmetry of genetic divergence are consistent with patterns of human gene flow.

q-bio.PE↗

arXiv2016

PipeCNN: An OpenCL-Based FPGA Accelerator for Large-Scale Convolution Neuron Networks

Convolutional neural networks (CNNs) have been widely employed in many applications such as image classification, video analysis and speech recognition. Being compute-intensive, CNN computations are mainly accelerated by GPUs with high power dissipations. Recently, studies were carried out exploiting FPGA as CNN accelerator because of its reconfigurability and energy efficiency advantage over GPU, especially when OpenCL-based high-level synthesis tools are now available providing fast verification and implementation flows. Previous OpenCL-based design only focused on creating a generic framework to identify performance-related hardware parameters, without utilizing FPGA's special capability of pipelining kernel functions to minimize memory bandwidth requirement. In this work, we propose an FPGA accelerator with a new architecture of deeply pipelined OpenCL kernels. Data reuse and task mapping techniques are also presented to improve design efficiency. The proposed schemes are verified by implementing two representative large-scale CNNs, AlexNet and VGG on Altera Stratix-V A7 FPGA. We have achieved a similar peak performance of 33.9 GOPS with a 34% resource reduction on DSP blocks compared to previous work. Our design is openly accessible and thus can be reused to explore new architectures for neural network accelerators.

cs.AR↗

arXiv2016

Wakefields studies for the SXFEL user facility

Besides the original seeded undulator line, in the Soft X-ray free-electron laser (SXFEL) user facility at Shanghai, a second undulator line based on self-amplified spontaneous emission is proposed to achieve 2 nm laser pulse with extremely high brightness. In this paper, the beam energy deviation induced by the undulator wakefields is numerically obtained, and it is verified to have a good agreement between 3D and 2D simulation results. The beam energy loss along the undulator degrades the expected FEL output performance. Impact of wakefields on pulse energy, radiation power and spectrum is discussed, as well as the benefits of compensation obtained with a taper in the undulator field. And using the planned SXFEL diagnostic, a longitudinal wakefields measurement experiment is proposed and simulated.

physics.acc-ph↗

arXiv2016

Chromatic Effect for THz Generation in a Novel Wave-front Tilt Scheme

Deriving single or few cycle terahertz pulse (THz) by intense femtosecond laser through cascaded optical rectification in electro-optic crystals is a crucial technique in cutting-edge time-resolved spectroscopy to characterize micro-scale structures and ultrafast dynamics. In the past decade, lithium niobate (LN) crystal implementation of wave-front tilt scheme has been prevalently used, while painstaking efforts have been invested in order to achieve higher THz conversion efficiency. In this research we developed a brand new type of LN crystal possessing dual-face-cut and Brewster coupling, and conducted experimental and simulative investigation systematically to optimize the multi-dimensionally entangled parameters in THz generation, predicting the extreme conversion efficiency of 10% is potentially promising at the THz absorption coefficient of 0.5cm-1. More remarkably, we first discovered that the chirp of the driving laser pulse plays a decisive role in the wave-front tilt scheme, and the THz generation efficiency could be enhanced tremendously by applying an appropriate chirp.

physics.optics↗

arXiv2016

An efficient threshold dynamics method for wetting on rough surfaces

The threshold dynamics method developed by Merriman, Bence and Osher (MBO) is an efficient method for simulating the motion by mean curvature flow when the interface is away from the solid boundary. Direct generalization of MBO-type methods to the wetting problem with interfaces intersecting the solid boundary is not easy because solving the heat equation in a general domain with a wetting boundary condition is not as efficient as it is with the original MBO method. The dynamics of the contact point also follows a different law compared with the dynamics of the interface away from the boundary. In this paper, we develop an efficient volume preserving threshold dynamics method for simulating wetting on rough surfaces. This method is based on minimization of the weighted surface area functional over an extended domain that includes the solid phase. The method is simple, stable with $O(N \log N)$ complexity per time step and is not sensitive to the inhomogeneity or roughness of the solid boundary.

math.NA↗

arXiv2016

System Combination for Short Utterance Speaker Recognition

For text-independent short-utterance speaker recognition (SUSR), the performance often degrades dramatically. This paper presents a combination approach to the SUSR tasks with two phonetic-aware systems: one is the DNN-based i-vector system and the other is our recently proposed subregion-based GMM-UBM system. The former employs phone posteriors to construct an i-vector model in which the shared statistics offers stronger robustness against limited test data, while the latter establishes a phone-dependent GMM-UBM system which represents speaker characteristics with more details. A score-level fusion is implemented to integrate the respective advantages from the two systems. Experimental results show that for the text-independent SUSR task, both the DNN-based i-vector system and the subregion-based GMM-UBM system outperform their respective baselines, and the score-level system combination delivers performance improvement.

cs.CL↗

arXiv2016

Multi-task Recurrent Model for Speech and Speaker Recognition

Although highly correlated, speech and speaker recognition have been regarded as two independent tasks and studied by two communities. This is certainly not the way that people behave: we decipher both speech content and speaker traits at the same time. This paper presents a unified model to perform speech and speaker recognition simultaneously and altogether. The model is based on a unified neural network where the output of one task is fed to the input of the other, leading to a multi-task recurrent network. Experiments show that the joint model outperforms the task-specific models on both the two tasks.

cs.CL↗

arXiv2016

Multi-task Recurrent Model for True Multilingual Speech Recognition

Research on multilingual speech recognition remains attractive yet challenging. Recent studies focus on learning shared structures under the multi-task paradigm, in particular a feature sharing structure. This approach has been found effective to improve performance on each individual language. However, this approach is only useful when the deployed system supports just one language. In a true multilingual scenario where multiple languages are allowed, performance will be significantly reduced due to the competition among languages in the decoding space. This paper presents a multi-task recurrent model that involves a multilingual speech recognition (ASR) component and a language recognition (LR) component, and the ASR component is informed of the language information by the LR component, leading to a language-aware recognition. We tested the approach on an English-Chinese bilingual recognition task. The results show that the proposed multi-task recurrent model can improve performance of multilingual recognition systems.

cs.CL↗

arXiv2016

OC16-CE80: A Chinese-English Mixlingual Database and A Speech Recognition Baseline

We present the OC16-CE80 Chinese-English mixlingual speech database which was released as a main resource for training, development and test for the Chinese-English mixlingual speech recognition (MixASR-CHEN) challenge on O-COCOSDA 2016. This database consists of 80 hours of speech signals recorded from more than 1,400 speakers, where the utterances are in Chinese but each involves one or several English words. Based on the database and another two free data resources (THCHS30 and the CMU dictionary), a speech recognition (ASR) baseline was constructed with the deep neural network-hidden Markov model (DNN-HMM) hybrid system. We then report the baseline results following the MixASR-CHEN evaluation rules and demonstrate that OC16-CE80 is a reasonable data resource for mixlingual research.

cs.CL↗

arXiv2016

Local Training for PLDA in Speaker Verification

PLDA is a popular normalization approach for the i-vector model, and it has delivered state-of-the-art performance in speaker verification. However, PLDA training requires a large amount of labeled development data, which is highly expensive in most cases. A possible approach to mitigate the problem is various unsupervised adaptation methods, which use unlabeled data to adapt the PLDA scattering matrices to the target domain. In this paper, we present a new `local training' approach that utilizes inaccurate but much cheaper local labels to train the PLDA model. These local labels discriminate speakers within a single conversion only, and so are much easier to obtain compared to the normal `global labels'. Our experiments show that the proposed approach can deliver significant performance improvement, particularly with limited globally-labeled data.

cs.SD↗

arXiv2016

AP16-OL7: A Multilingual Database for Oriental Languages and A Language Recognition Baseline

We present the AP16-OL7 database which was released as the training and test data for the oriental language recognition (OLR) challenge on APSIPA 2016. Based on the database, a baseline system was constructed on the basis of the i-vector model. We report the baseline results evaluated in various metrics defined by the AP16-OLR evaluation plan and demonstrate that AP16-OL7 is a reasonable data resource for multilingual research.

cs.CL↗

arXiv2016

Deflecting cavity dynamics for time-resolved machine studies of SXFEL user facility

Radio frequency deflectors are widely used for time-resolved electron beam energy, emittance and radiation profile measurements in modern free electron laser facilities. In this paper, we present the beam dynamics aspects of the deflecting cavity of SXFEL user facility, which is located at the exit of the undulator. With a targeted time resolution around 10 fs, it is expected to be an important tool for time-resolved commissioning and machine studies for SXFEL user facility.

physics.acc-ph↗

arXiv2016

Generating coherent soft x-ray pulses in the water window with a high-brightness seeded free-electron laser

We propose a new scheme to generate high-brightness and temporal coherent soft x-ray radiation in a seeded free-electron laser. The proposed scheme is based the coherent harmonic generation (CHG) and superradiant principles. A CHG scheme is first used to generate coherent signal at ultra-high harmonics of the seed. This coherent signal is then amplified by a series of chicane-undulator modules via the fresh bunch and superradiant processes in the following radiator. Using a representative of realistic set of parameters, three-dimensional simulations have been carried out and the simulations results demonstrated that 10 GW-level ultra-short coherent radiation pulses in the water window can be achieved by using the proposed technique.

physics.acc-ph↗

arXiv2016

Nonintersecting Brownian bridges between reflecting or absorbing walls

We study a model of nonintersecting Brownian bridges on an interval with either absorbing or reflecting walls at the boundaries, focusing on the point in space-time at which the particles meet the wall. These processes are determinantal, and in different scaling limits when the particles approach the reflecting (resp. absorbing) walls we obtain hard-edge limiting kernels which are the even (resp. odd) parts of the Pearcey and tacnode kernels. We also show that in the single time case, our hard-edge tacnode kernels are equivalent to the ones studied by Delvaux [16], defined in terms of a $4\times 4$ Lax pair for the inhomogeneous Painlevé II equation (PII). As a technical ingredient in the proof, we construct a Schlesinger transform for the $4 \times 4$ Lax pair in [16] which preserves the Hastings--McLeod solutions to PII.

math.PR↗

arXiv2016

Propagation of singular behavior for Gaussian perturbations of random matrices

We study the asymptotic behavior of the eigenvalues of Gaussian perturbations of large Hermitian random matrices for which the limiting eigenvalue density vanishes at a singular interior point or vanishes faster than a square root at a singular edge point. First, we show that the singular behavior propagates macroscopically for sufficiently small Gaussian perturbations, and we describe the macroscopic eigenvalue behavior for Gaussian perturbations of critical size. Secondly, for sufficiently small Gaussian perturbations of unitary invariant random matrices, we prove that the microscopic eigenvalue correlations near the singular point are described by the same limiting kernel as in the unperturbed case. We also interpret our results in terms of nonintersecting Brownian paths with random starting positions, and we establish multi-time generalizations of the microscopic results.

math.PR↗

arXiv2016

Two Lax systems for the Painlevé II equation, and two related kernels in random matrix theory

We consider two Lax systems for the homogeneous Painlevé II equation: one of size $2\times 2$ studied by Flaschka and Newell in the early 1980's, and one of size $4\times 4$ introduced by Delvaux-Kuijlaars-Zhang and Duits-Geudens in the early 2010's. We prove that solutions to the $4\times 4$ system can be derived from those to the $2\times 2$ system via an integral transform, and consequently relate the Stokes multipliers for the two systems. As corollaries we are able to express two kernels for determinantal processes as contour integrals involving the Flaschka-Newell Lax system: the tacnode kernel arising in models of nonintersecting paths, and a critical kernel arising in a two-matrix model.

nlin.SI↗

arXiv2016

An efficient iterative thresholding method for image segmentation

We proposed an efficient iterative thresholding method for multi-phase image segmentation. The algorithm is based on minimizing piecewise constant Mumford-Shah functional in which the contour length (or perimeter) is approximated by a non-local multi-phase energy. The minimization problem is solved by an iterative method. Each iteration consists of computing simple convolutions followed by a thresholding step. The algorithm is easy to implement and has the optimal complexity $O(N \log N)$ per iteration. We also show that the iterative algorithm has the total energy decaying property. We present some numerical results to show the efficiency of our method.

cs.CV↗

arXiv2016

Chinese Song Iambics Generation with Neural Attention-based Model

Learning and generating Chinese poems is a charming yet challenging task. Traditional approaches involve various language modeling and machine translation techniques, however, they perform not as well when generating poems with complex pattern constraints, for example Song iambics, a famous type of poems that involve variable-length sentences and strict rhythmic patterns. This paper applies the attention-based sequence-to-sequence model to generate Chinese Song iambics. Specifically, we encode the cue sentences by a bi-directional Long-Short Term Memory (LSTM) model and then predict the entire iambic with the information provided by the encoder, in the form of an attention-based LSTM that can regularize the generation process by the fine structure of the input cues. Several techniques are investigated to improve the model, including global context integration, hybrid style training, character vector initialization and adaptation. Both the automatic and subjective evaluation results show that our model indeed can learn the complex structural and rhythmic patterns of Song iambics, and the generation is rather successful.

cs.CL↗