Skip to main content

quốc Nguyễn

Followers

2

Following

1

Public Views

Makhlouf Derdour

Universite de Tebessa

Núria Busom Figueres

Mohammad Shojafar

University of Surrey

Amalia Pallares

University of Illinois at Chicago

Pawani Porambage

University of Moratuwa

Md Zakirul Alam Bhuiyan

University of Calcutta

Maria Celestin Vigila S

Interests

Uploads

Papers by quốc Nguyễn

Crystal: A Privacy-Preserving Distributed Reputation Management

Natural Language Processing, 2021

This paper investigates the situation in which exists the unshared Internet in specific areas whi... more This paper investigates the situation in which exists the unshared Internet in specific areas while users in there need instant advice from others nearby. Hence, a peer-to-peer network is necessary and established by connecting all neighbouring mobile devices so that they can exchange questions and recommendations. However, not all received recommendations are reliable as users may be unknown to each other. Therefore, the trustworthiness of advice is evaluated based on the advisor's reputation score. The reputation score is locally stored in the user’s mobile device. It is not completely guaranteed that the reputation score is trustful if its owner uses it for a wrong intention. In addition, another privacy problem is about honestly auditing the reputation score on the advising user by the questioning user. Therefore, this work proposes a security model, namely Crystal, for securely managing distributed reputation scores and for preserving user privacy. Crystal ensures that the ...

Comparison of forecast models of production of dairy cows combining animal and diet parameters

Computers and Electronics in Agriculture, 2020

We study the effect of nutritional diet characteristics on the lactating Holstein-Friesian dairy ... more We study the effect of nutritional diet characteristics on the lactating Holstein-Friesian dairy cows in Brittany, France from 36 individuals. An analysis of the relations between fat/protein content and milk yield was implemented for our dataset. The fat and protein production increase at a slower rate as milk yield increases. The importance of chemical composition on milk production is studied using the linear model. The data analysis confirms the importance of Starch, crude fiber, and protein which have a positive effect on milk production. This analysis also confirms the previous study on the effect of parity on the production. After that, the milk production forecasting is investigated using both linear models and machine learning approaches (support vector machine, random forest, neural network). We study the performance of multiple linear regression and machine learning-based models in both non-autoregressive and autoregressive cases at the individual level. The autoregressive models, which take into account the previously observed milk yield, have proven to significantly outperform the non-autoregressive approaches. Moreover, the computational cost of each approach is presented in the paper. While the random forest algorithm gives the best performance in both non-autoregressive and autoregressive approaches. The support vector machine algorithm gives a very close performance with a substantial less computing time. The support vector machine

On the performance of VSI Shewhart control chart for monitoring the coefficient of variation in the presence of measurement errors

The International Journal of Advanced Manufacturing Technology, 2019

In this paper, we propose a variable sampling interval Shewhart control chart to monitor the coef... more In this paper, we propose a variable sampling interval Shewhart control chart to monitor the coefficient of variation (CV) squared, denoted by VSI SH-γ 2. The new model overcomes the ARL-biased (average run length) property of the control chart monitoring the CV in a previous study by designing two one-sided charts rather than one two-sided chart. Moreover, the effect of measurement error on the performance of the VSI SH-γ 2 control chart is investigated. The incorrect formula for the distribution of the CV in the presence of measurement error in a former study is fixed. Numerical simulations show that the precision errors and accuracy errors do have negative influences on the VSI SH-γ 2 chart. An appropriate strategy based on the obtained results is suggested to reduce these negative effects.

Portion size selection as related to product and consumer characteristics studied by PLS path modelling

Food Quality and Preference, 2018

Expectations of satiation and satiety have been increasingly investigated because of 12 the inter... more Expectations of satiation and satiety have been increasingly investigated because of 12 the interest in how they, along with liking, can modulate portion-size selection. 13 Consumer characteristics can also be important when consumers select their portion 14 size. However, the contribution and interaction of consumers and product aspects to 15 portion size selection has not been unveiled. This study aims to better understanding 16 these complex relations by simultaneously assessing the relative influence of 17 consumer characteristics and product related properties on portion size selection 18 utilizing PLS-Path Modelling (PLS-PM) approach. 19 In this study, consumers (n=101) answered questions regarding attitudes to health and 20 hedonic characteristics of foods, and completed hunger and fullness questions. In an 21 evaluation step, they tasted eight samples of yogurt with different textures and rated 22 liking, expected satiation, expected satiety and portion size. The consumers were also 23 classified on their mouth behaviour by using the JBMB™ tool. 24 Results showed that liking, satiation, satiety and portion size depended firstly on the 25 thickness, and then on the particle size of samples. PLS-PM was used to generate a 26 model, indicating that liking was a direct predictor of portion size, with a stronger effect 27 than satiation or satiety. The relationship between liking and satiety was observed both 28 in direct direction (liking-satiety) and also indirect direction throughout satiation (liking-29 satiation-satiety). The former was negative effect and the latter was positive effect 30 depending on the criteria which consumers used. 31 These findings implied that liking is a main factor in the prediction of portion size 32 however the relations are complex.

The impact of (68)Ga-PSMA PET/CT on management intent in prostate cancer: results of an Australian prospective multicenter study

Journal of nuclear medicine : official publication, Society of Nuclear Medicine, Jan 23, 2017

scanning has been shown to be more sensitive than conventional imaging techniques in patients wit... more scanning has been shown to be more sensitive than conventional imaging techniques in patients with prostate cancer. This prospective Australian multicenter study assessed whether 68 Ga-PSMA PET/CT imaging affects management intent in patients with primary or recurrent prostate cancer. Methods: Before undertaking 68 Ga-PSMA PET imaging, referring medical specialists completed a questionnaire detailing relevant demographic and clinical data as well as their proposed management plan. A separate followup questionnaire was completed after the 68 Ga-PSMA PET/CT scan results were available to determine whether the management plan would change. Results: A total of 431 patients with prostate cancer from 4 Australian centers had pre-and post-68 Ga-PSMA management plans completed. Scans were obtained for primary staging of intermediate-and high-risk disease in 25% of patients and for restaging/biochemical recurrence in 75% of patients. Overall, 68 Ga-PSMA PET/CT scanning led to a change in planned management in 51% of patients. The impact was greater in the group of patients with biochemical failure after definitive surgery or radiation treatment (62% change in management intent) than in patients undergoing primary staging (21% change). Imaging with 68 Ga-PSMA PET/CT revealed unsuspected disease in the prostate bed in 27% of patients, locoregional lymph nodes in 39%, and distant metastatic disease in 16%. Conclusion: 68 Ga-PSMA PET/CT scans detect previously unsuspected disease and may influence planned clinical management in a high proportion of patients with prostate cancer. The impact was greater in patients with biochemical recurrence. These results demonstrate the potential clinical value of 68 Ga-PSMA PET/CT in management of prostate cancer.

Discreteness of interior transmission eigenvalues revisited

Calculus of Variations and Partial Differential Equations, 2017

This paper is devoted to the discreteness of the transmission eigenvalue problems. It is known th... more This paper is devoted to the discreteness of the transmission eigenvalue problems. It is known that this problem is not self-adjoint and a priori estimates are non-standard and do not hold in general. Two approaches are used. The first one is based on the multiplier technique and the second one is based on the Fourier analysis. The key point of the analysis is to establish the compactness and the uniqueness for Cauchy problems under various conditions. Using these approaches, we are able to rediscover quite a few known discreteness results in the literature and obtain various new results for which only the information near the boundary are required and there might be no contrast of the coefficients on the boundary.

FEM-based discretization-invariant MCMC methods for PDE-constrained Bayesian inverse problems

Inverse Problems and Imaging, 2016

We present a systematic construction of FEM-based dimension-independent (discretization-invariant... more We present a systematic construction of FEM-based dimension-independent (discretization-invariant) Markov chain Monte Carlo (MCMC) approaches to explore PDE-constrained Bayesian inverse problems in infinite dimensional parameter spaces. In particular, we consider two frameworks to achieve this goal: Metropolize-then-discretize and discretize-then-Metropolize. The former refers to the method of discretizing function-space MCMC methods. The latter, on the other hand, first discretizes the Bayesian inverse problem and then proposes MCMC methods for the resulting discretized posterior probability density. In general, these two frameworks do not commute, that is, the resulting finite dimensional MCMC algorithms are not identical. The discretization step of the former may not be trivial since it involves both numerical analysis and probability theory, while the latter, perhaps "easier", may not be discretization-invariant using traditional approaches. This paper constructively develops finite element (FEM) discretization schemes for both frameworks and shows that both commutativity and discretization-invariant are attained. In particular, it shows how to construct discretize-then-Metropolize approaches for both Metropolis-adjusted Langevin algorithm and hybrid Monte Carlo method that commute with their Metropolize-then-discretize counterparts. The key that enables this achievement is a proper FEM discretization of the prior, the likelihood, and the Bayes' formula, together with a correct definition of quantities such as gradient and covariance matrix in discretized finite dimensional parameter spaces. The implication is that practitioners can take advantage of the developments in this paper to straightforwardly construct discretization-invariant

Stability properties for quasilinear parabolic equations with measure data

Journal of the European Mathematical Society, 2015

where p > 1, µ ∈ M b (Q) and u 0 ∈ L 1 (). Our main result is a stability theorem extending the r... more

Global estimates for quasilinear parabolic equations on Reifenberg flat domains and its applications to Riccati type parabolic equations with distributional data

Calculus of Variations and Partial Differential Equations, 2015

In this paper, we prove global weighted Lorentz and Lorentz-Morrey estimates for gradients of sol... more In this paper, we prove global weighted Lorentz and Lorentz-Morrey estimates for gradients of solutions to the quasilinear parabolic equations: ut − div(A(x, t, ∇u)) = div(F), in a bounded domain Ω × (0, T) ⊂ R N+1 , under minimal regularity assumptions on the boundary of domain and on nonlinearity A. Then results yields existence of a solution to the Riccati type parabolic equations: ut − div(A(x, t, ∇u)) = |∇u| q + div(F) + µ, where q > 1 and µ is a bounded Radon measure.

Optical coherence tomography using the Niris system in otolaryngology

SPIE Proceedings, 2009

Objectives: To determine the feasibility and accuracy of the Niris Optical Coherence Tomography (... more Objectives: To determine the feasibility and accuracy of the Niris Optical Coherence Tomography (OCT) system in imaging of the mucosal abnormalities of the head and neck. The Niris system is the first commercially available OCT device for applications outside ophthalmology. Methods: We obtained OCT images of benign, premalignant and malignant lesions throughout the head and neck, using the Niris OCT imaging system (Imalux, Cleveland, OH). This imaging system has a tissue penetration depth of approximately 1-2mm, a scanning range of 2mm and a spatial depth resolution of approximately 10-20μm. Imaging was performed in the outpatient setting and in the operating room using a flexible probe. Results: High-resolution cross-sectional images from the oral cavity, nasal cavity, ears and larynx showed distinct layers and structures such as mucosa layer, basal membrane and lamina propria, were clearly identified. In the pathology images disruption of the basal membrane was clearly shown. Device setup took approximately 5 minutes and the image acquisition was rapid. The system can be operated by the person performing the exam. Conclusions: The Niris system is non invasive and easy to incorporate into the operating room and the clinic. It requires minimal setup and requires only one person to operate. The unique ability of the OCT offers high-resolution images showing the microanatomy of different sites. OCT imaging with the Niris device potentially offers an efficient, quick and reliable imaging modality in guiding surgical biopsies, intra-operative decision

Active vibration control of an axially moving beam using varying velocity method

Modeling of materials capable of solid–solid phase transformation. Application to the analytical solution of the semi-infinite mode III crack problem in a phase-changing solid

International Journal of Non-Linear Mechanics, 2015

The paper proposes two frameworks for the derivation of constitutive models for solids undergoing... more The paper proposes two frameworks for the derivation of constitutive models for solids undergoing phase transformations. Two distinct approaches are considered: the first is based on the assumption that solid phases within the material are finely mixed whereas the second considers the material as a heterogeneous solution of different phase fragments and uses the homogenization theory to derive constitutive relations at the macroscale. For both approaches, the mathematical representation of the material behavior is intentionally kept simple and the derivations are fully developed for ease of future modification and use by the readership. It is shown that in the case of reversible phase transformation, the energy of the material can be obtained by convexifiying the energy functions of the constituent phases. It is further shown that for dissipative phase transformation the material behavior can be made stable by deriving the evolution equations of the state variables from adequately chosen dissipation potentials. Some new results of existence and uniqueness of the solution of boundary value problems of structures undergoing phase change are also given. As an application, a schematic model is derived and used to obtain analytical solutions for the problem of semi-infinite mode III crack in a solid capable of phase transformation.

$Fig. 2. Distribution in the xy plane of the volume fraction of martensite. where p is the hydrostatic pressure and o’ is the stress deviator. If the volume fractions z; for ie {1,2,3} are strictly positive, the expressions of the thermodynamic forces A; = V“+v =0 lead to$

Variational Principles and Variational Inequalities

Fluid Mechanics of Viscoplasticity, 2015

In this chapter, we shall derive twoVariational principle variational principles and variational ... more

DNN acoustic modeling with modular multi-lingual feature extraction networks

In this work, we propose several deep neural network architectures that are able to leverage data... more In this work, we propose several deep neural network architectures that are able to leverage data from multiple languages. Modularity is achieved by training networks for extracting high-level features and for estimating phoneme state posteriors separately, and then combining them for decoding in a hybrid DNN/HMM setup. This approach has been shown to achieve superior performance for single-language systems, and here we demonstrate that feature extractors benefit significantly from being trained as multilingual networks with shared hidden representations. We also show that existing mono-lingual networks can be re-used in a modular fashion to achieve a similar level of performance without having to train new networks on multilingual data. Furthermore, we investigate in extending these architectures to make use of language-specific acoustic features. Evaluations are performed on a low-resource conversational telephone speech transcription task in Vietnamese, while additional data for acoustic model training is provided in Pashto, Tagalog, Turkish, and Cantonese. Improvements of up to 17.4% and 13.8% over mono-lingual GMMs and DNNs, respectively, are obtained.

Wiener Criteria for Existence of Large Solutions of Quasilinear Elliptic Equations with Absorption

Potential Analysis, 2014

We obtain sufficient conditions, expressed in terms of Wiener type tests involving Hausdorff or B... more We obtain sufficient conditions, expressed in terms of Wiener type tests involving Hausdorff or Bessel capacities, for the existence of large solutions to equations (1) −∆ p u + e u − 1 = 0 or (2) −∆ p u + u q = 0 in a bounded domain Ω when q > p − 1 > 0. We apply our results to equations (3) −∆ p u + a |∇u| q + bu s = 0, (4) ∆ p u + u −γ = 0 with 1 < p ≤ 2, 1 ≤ q ≤ p, a > 0, b > 0 and q > p − 1, s ≥ p − 1, γ > 0.

Multi features models for robust lip tracking

2008 10th International Conference on Control, Automation, Robotics and Vision, 2008

We propose and evaluate a novel method for enhancing performance of lips contour tracking, which ... more We propose and evaluate a novel method for enhancing performance of lips contour tracking, which is based on the concept of statistic shape models (ASM) and multi features. On the first image of the video sequence, lip region is detected using the Bayesian&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;s rule in which lip color information is modeled by using the Gaussian mixture model (GMM) and the

Adaptive boundary control of an axially moving string system: Application to container cranes

2009 IEEE International Symposium on Industrial Electronics, 2009

This paper proposes an adaptive boundary control to an axially moving string system, which couple... more This paper proposes an adaptive boundary control to an axially moving string system, which couples with a mass-damper-spring (MDS) controller at its right-hand-side (RHS) boundary. Unknown parameters appearing in the system equation are assumed constant and estimated on-line by using adaptation laws. The adaptive computed-torque control algorithm applied to robot manipulators of lumped systems is extended to design the adaptive boundary controller for the coupling system. It is found that the control force and update laws depend only on the displacement, velocity and slope of the string at the RHS boundary. Lyapunov stability guarantees the convergence of the tracking error to zero. Finally, the performance of the proposed controller is demonstrated by numerical simulations.

Models of tone for tonal and non-tonal languages

2013 IEEE Workshop on Automatic Speech Recognition and Understanding, 2013

Conventional wisdom in automatic speech recognition asserts that pitch information is not helpful... more Conventional wisdom in automatic speech recognition asserts that pitch information is not helpful in building speech recognizers for non-tonal languages and contributes only modestly to performance in speech recognizers for tonal languages. To maintain consistency between different systems, pitch is therefore often ignored, trading the slight performance benefits for greater system uniformity/ simplicity. In this paper, we report results that challenge this conventional approach. We present new models of tone that deliver consistent performance improvements for tonal languages (Cantonese, Vietnamese) and even modest improvements for non-tonal languages. Using neural networks for feature integration and fusion, these models achieve significant gains throughout, and provide us with system uniformity and standardization across all languages, tonal and non-tonal.

Multigrid Optimization Schemes for Solving Bose–Einstein Condensate Control Problems

SIAM Journal on Scientific Computing, 2008

The control of the transport of Bose-Einstein condensates in magnetic microtraps is formulated wi... more The control of the transport of Bose-Einstein condensates in magnetic microtraps is formulated within the framework of optimal control theory and solved by multigrid optimization schemes. The time evolution of the wave function of Bose-Einstein condensates is governed by the Gross-Pitaevskii equation and can be manipulated through variation of a controllable magnetic confinement potential. In order to define an optimal control strategy, an appropriate cost functional is introduced that must be minimized under the constraint given by the dynamic equation. The resulting optimality system consists of two nonlinear Schrödinger-type equations with opposite time orientation coupled with an elliptic equation for the control function. These equations are approximated by using a time-splitting pseudospectral method and finite differences. To solve the resulting problem a cascadic nonlinear conjugate gradient scheme and a multigrid optimization scheme are considered. The convergence properties of these two schemes are investigated theoretically, and their computational performance is discussed based on results of numerical experiments. It appears that the multigrid optimization scheme provides a robust optimization strategy.

An elliptic semilinear equation with source term involving boundary measures: the subcritical case

Revista Matemática Iberoamericana, 2000

Crystal: A Privacy-Preserving Distributed Reputation Management

Natural Language Processing, 2021

This paper investigates the situation in which exists the unshared Internet in specific areas whi... more This paper investigates the situation in which exists the unshared Internet in specific areas while users in there need instant advice from others nearby. Hence, a peer-to-peer network is necessary and established by connecting all neighbouring mobile devices so that they can exchange questions and recommendations. However, not all received recommendations are reliable as users may be unknown to each other. Therefore, the trustworthiness of advice is evaluated based on the advisor's reputation score. The reputation score is locally stored in the user’s mobile device. It is not completely guaranteed that the reputation score is trustful if its owner uses it for a wrong intention. In addition, another privacy problem is about honestly auditing the reputation score on the advising user by the questioning user. Therefore, this work proposes a security model, namely Crystal, for securely managing distributed reputation scores and for preserving user privacy. Crystal ensures that the ...

Comparison of forecast models of production of dairy cows combining animal and diet parameters

Computers and Electronics in Agriculture, 2020

We study the effect of nutritional diet characteristics on the lactating Holstein-Friesian dairy ... more We study the effect of nutritional diet characteristics on the lactating Holstein-Friesian dairy cows in Brittany, France from 36 individuals. An analysis of the relations between fat/protein content and milk yield was implemented for our dataset. The fat and protein production increase at a slower rate as milk yield increases. The importance of chemical composition on milk production is studied using the linear model. The data analysis confirms the importance of Starch, crude fiber, and protein which have a positive effect on milk production. This analysis also confirms the previous study on the effect of parity on the production. After that, the milk production forecasting is investigated using both linear models and machine learning approaches (support vector machine, random forest, neural network). We study the performance of multiple linear regression and machine learning-based models in both non-autoregressive and autoregressive cases at the individual level. The autoregressive models, which take into account the previously observed milk yield, have proven to significantly outperform the non-autoregressive approaches. Moreover, the computational cost of each approach is presented in the paper. While the random forest algorithm gives the best performance in both non-autoregressive and autoregressive approaches. The support vector machine algorithm gives a very close performance with a substantial less computing time. The support vector machine

On the performance of VSI Shewhart control chart for monitoring the coefficient of variation in the presence of measurement errors

The International Journal of Advanced Manufacturing Technology, 2019

In this paper, we propose a variable sampling interval Shewhart control chart to monitor the coef... more In this paper, we propose a variable sampling interval Shewhart control chart to monitor the coefficient of variation (CV) squared, denoted by VSI SH-γ 2. The new model overcomes the ARL-biased (average run length) property of the control chart monitoring the CV in a previous study by designing two one-sided charts rather than one two-sided chart. Moreover, the effect of measurement error on the performance of the VSI SH-γ 2 control chart is investigated. The incorrect formula for the distribution of the CV in the presence of measurement error in a former study is fixed. Numerical simulations show that the precision errors and accuracy errors do have negative influences on the VSI SH-γ 2 chart. An appropriate strategy based on the obtained results is suggested to reduce these negative effects.

Portion size selection as related to product and consumer characteristics studied by PLS path modelling

Food Quality and Preference, 2018

Expectations of satiation and satiety have been increasingly investigated because of 12 the inter... more Expectations of satiation and satiety have been increasingly investigated because of 12 the interest in how they, along with liking, can modulate portion-size selection. 13 Consumer characteristics can also be important when consumers select their portion 14 size. However, the contribution and interaction of consumers and product aspects to 15 portion size selection has not been unveiled. This study aims to better understanding 16 these complex relations by simultaneously assessing the relative influence of 17 consumer characteristics and product related properties on portion size selection 18 utilizing PLS-Path Modelling (PLS-PM) approach. 19 In this study, consumers (n=101) answered questions regarding attitudes to health and 20 hedonic characteristics of foods, and completed hunger and fullness questions. In an 21 evaluation step, they tasted eight samples of yogurt with different textures and rated 22 liking, expected satiation, expected satiety and portion size. The consumers were also 23 classified on their mouth behaviour by using the JBMB™ tool. 24 Results showed that liking, satiation, satiety and portion size depended firstly on the 25 thickness, and then on the particle size of samples. PLS-PM was used to generate a 26 model, indicating that liking was a direct predictor of portion size, with a stronger effect 27 than satiation or satiety. The relationship between liking and satiety was observed both 28 in direct direction (liking-satiety) and also indirect direction throughout satiation (liking-29 satiation-satiety). The former was negative effect and the latter was positive effect 30 depending on the criteria which consumers used. 31 These findings implied that liking is a main factor in the prediction of portion size 32 however the relations are complex.

The impact of (68)Ga-PSMA PET/CT on management intent in prostate cancer: results of an Australian prospective multicenter study

Journal of nuclear medicine : official publication, Society of Nuclear Medicine, Jan 23, 2017

scanning has been shown to be more sensitive than conventional imaging techniques in patients wit... more scanning has been shown to be more sensitive than conventional imaging techniques in patients with prostate cancer. This prospective Australian multicenter study assessed whether 68 Ga-PSMA PET/CT imaging affects management intent in patients with primary or recurrent prostate cancer. Methods: Before undertaking 68 Ga-PSMA PET imaging, referring medical specialists completed a questionnaire detailing relevant demographic and clinical data as well as their proposed management plan. A separate followup questionnaire was completed after the 68 Ga-PSMA PET/CT scan results were available to determine whether the management plan would change. Results: A total of 431 patients with prostate cancer from 4 Australian centers had pre-and post-68 Ga-PSMA management plans completed. Scans were obtained for primary staging of intermediate-and high-risk disease in 25% of patients and for restaging/biochemical recurrence in 75% of patients. Overall, 68 Ga-PSMA PET/CT scanning led to a change in planned management in 51% of patients. The impact was greater in the group of patients with biochemical failure after definitive surgery or radiation treatment (62% change in management intent) than in patients undergoing primary staging (21% change). Imaging with 68 Ga-PSMA PET/CT revealed unsuspected disease in the prostate bed in 27% of patients, locoregional lymph nodes in 39%, and distant metastatic disease in 16%. Conclusion: 68 Ga-PSMA PET/CT scans detect previously unsuspected disease and may influence planned clinical management in a high proportion of patients with prostate cancer. The impact was greater in patients with biochemical recurrence. These results demonstrate the potential clinical value of 68 Ga-PSMA PET/CT in management of prostate cancer.

Discreteness of interior transmission eigenvalues revisited

Calculus of Variations and Partial Differential Equations, 2017

This paper is devoted to the discreteness of the transmission eigenvalue problems. It is known th... more This paper is devoted to the discreteness of the transmission eigenvalue problems. It is known that this problem is not self-adjoint and a priori estimates are non-standard and do not hold in general. Two approaches are used. The first one is based on the multiplier technique and the second one is based on the Fourier analysis. The key point of the analysis is to establish the compactness and the uniqueness for Cauchy problems under various conditions. Using these approaches, we are able to rediscover quite a few known discreteness results in the literature and obtain various new results for which only the information near the boundary are required and there might be no contrast of the coefficients on the boundary.

FEM-based discretization-invariant MCMC methods for PDE-constrained Bayesian inverse problems

Inverse Problems and Imaging, 2016

We present a systematic construction of FEM-based dimension-independent (discretization-invariant... more We present a systematic construction of FEM-based dimension-independent (discretization-invariant) Markov chain Monte Carlo (MCMC) approaches to explore PDE-constrained Bayesian inverse problems in infinite dimensional parameter spaces. In particular, we consider two frameworks to achieve this goal: Metropolize-then-discretize and discretize-then-Metropolize. The former refers to the method of discretizing function-space MCMC methods. The latter, on the other hand, first discretizes the Bayesian inverse problem and then proposes MCMC methods for the resulting discretized posterior probability density. In general, these two frameworks do not commute, that is, the resulting finite dimensional MCMC algorithms are not identical. The discretization step of the former may not be trivial since it involves both numerical analysis and probability theory, while the latter, perhaps "easier", may not be discretization-invariant using traditional approaches. This paper constructively develops finite element (FEM) discretization schemes for both frameworks and shows that both commutativity and discretization-invariant are attained. In particular, it shows how to construct discretize-then-Metropolize approaches for both Metropolis-adjusted Langevin algorithm and hybrid Monte Carlo method that commute with their Metropolize-then-discretize counterparts. The key that enables this achievement is a proper FEM discretization of the prior, the likelihood, and the Bayes' formula, together with a correct definition of quantities such as gradient and covariance matrix in discretized finite dimensional parameter spaces. The implication is that practitioners can take advantage of the developments in this paper to straightforwardly construct discretization-invariant

Stability properties for quasilinear parabolic equations with measure data

Journal of the European Mathematical Society, 2015

where p > 1, µ ∈ M b (Q) and u 0 ∈ L 1 (). Our main result is a stability theorem extending the r... more

Global estimates for quasilinear parabolic equations on Reifenberg flat domains and its applications to Riccati type parabolic equations with distributional data

Calculus of Variations and Partial Differential Equations, 2015

In this paper, we prove global weighted Lorentz and Lorentz-Morrey estimates for gradients of sol... more In this paper, we prove global weighted Lorentz and Lorentz-Morrey estimates for gradients of solutions to the quasilinear parabolic equations: ut − div(A(x, t, ∇u)) = div(F), in a bounded domain Ω × (0, T) ⊂ R N+1 , under minimal regularity assumptions on the boundary of domain and on nonlinearity A. Then results yields existence of a solution to the Riccati type parabolic equations: ut − div(A(x, t, ∇u)) = |∇u| q + div(F) + µ, where q > 1 and µ is a bounded Radon measure.

Optical coherence tomography using the Niris system in otolaryngology

SPIE Proceedings, 2009

Objectives: To determine the feasibility and accuracy of the Niris Optical Coherence Tomography (... more Objectives: To determine the feasibility and accuracy of the Niris Optical Coherence Tomography (OCT) system in imaging of the mucosal abnormalities of the head and neck. The Niris system is the first commercially available OCT device for applications outside ophthalmology. Methods: We obtained OCT images of benign, premalignant and malignant lesions throughout the head and neck, using the Niris OCT imaging system (Imalux, Cleveland, OH). This imaging system has a tissue penetration depth of approximately 1-2mm, a scanning range of 2mm and a spatial depth resolution of approximately 10-20μm. Imaging was performed in the outpatient setting and in the operating room using a flexible probe. Results: High-resolution cross-sectional images from the oral cavity, nasal cavity, ears and larynx showed distinct layers and structures such as mucosa layer, basal membrane and lamina propria, were clearly identified. In the pathology images disruption of the basal membrane was clearly shown. Device setup took approximately 5 minutes and the image acquisition was rapid. The system can be operated by the person performing the exam. Conclusions: The Niris system is non invasive and easy to incorporate into the operating room and the clinic. It requires minimal setup and requires only one person to operate. The unique ability of the OCT offers high-resolution images showing the microanatomy of different sites. OCT imaging with the Niris device potentially offers an efficient, quick and reliable imaging modality in guiding surgical biopsies, intra-operative decision

Active vibration control of an axially moving beam using varying velocity method

Modeling of materials capable of solid–solid phase transformation. Application to the analytical solution of the semi-infinite mode III crack problem in a phase-changing solid

International Journal of Non-Linear Mechanics, 2015

The paper proposes two frameworks for the derivation of constitutive models for solids undergoing... more The paper proposes two frameworks for the derivation of constitutive models for solids undergoing phase transformations. Two distinct approaches are considered: the first is based on the assumption that solid phases within the material are finely mixed whereas the second considers the material as a heterogeneous solution of different phase fragments and uses the homogenization theory to derive constitutive relations at the macroscale. For both approaches, the mathematical representation of the material behavior is intentionally kept simple and the derivations are fully developed for ease of future modification and use by the readership. It is shown that in the case of reversible phase transformation, the energy of the material can be obtained by convexifiying the energy functions of the constituent phases. It is further shown that for dissipative phase transformation the material behavior can be made stable by deriving the evolution equations of the state variables from adequately chosen dissipation potentials. Some new results of existence and uniqueness of the solution of boundary value problems of structures undergoing phase change are also given. As an application, a schematic model is derived and used to obtain analytical solutions for the problem of semi-infinite mode III crack in a solid capable of phase transformation.

$Fig. 2. Distribution in the xy plane of the volume fraction of martensite. where p is the hydrostatic pressure and o’ is the stress deviator. If the volume fractions z; for ie {1,2,3} are strictly positive, the expressions of the thermodynamic forces A; = V“+v =0 lead to$

Variational Principles and Variational Inequalities

Fluid Mechanics of Viscoplasticity, 2015

In this chapter, we shall derive twoVariational principle variational principles and variational ... more

DNN acoustic modeling with modular multi-lingual feature extraction networks

In this work, we propose several deep neural network architectures that are able to leverage data... more In this work, we propose several deep neural network architectures that are able to leverage data from multiple languages. Modularity is achieved by training networks for extracting high-level features and for estimating phoneme state posteriors separately, and then combining them for decoding in a hybrid DNN/HMM setup. This approach has been shown to achieve superior performance for single-language systems, and here we demonstrate that feature extractors benefit significantly from being trained as multilingual networks with shared hidden representations. We also show that existing mono-lingual networks can be re-used in a modular fashion to achieve a similar level of performance without having to train new networks on multilingual data. Furthermore, we investigate in extending these architectures to make use of language-specific acoustic features. Evaluations are performed on a low-resource conversational telephone speech transcription task in Vietnamese, while additional data for acoustic model training is provided in Pashto, Tagalog, Turkish, and Cantonese. Improvements of up to 17.4% and 13.8% over mono-lingual GMMs and DNNs, respectively, are obtained.

Wiener Criteria for Existence of Large Solutions of Quasilinear Elliptic Equations with Absorption

Potential Analysis, 2014

We obtain sufficient conditions, expressed in terms of Wiener type tests involving Hausdorff or B... more We obtain sufficient conditions, expressed in terms of Wiener type tests involving Hausdorff or Bessel capacities, for the existence of large solutions to equations (1) −∆ p u + e u − 1 = 0 or (2) −∆ p u + u q = 0 in a bounded domain Ω when q > p − 1 > 0. We apply our results to equations (3) −∆ p u + a |∇u| q + bu s = 0, (4) ∆ p u + u −γ = 0 with 1 < p ≤ 2, 1 ≤ q ≤ p, a > 0, b > 0 and q > p − 1, s ≥ p − 1, γ > 0.

Multi features models for robust lip tracking

2008 10th International Conference on Control, Automation, Robotics and Vision, 2008

We propose and evaluate a novel method for enhancing performance of lips contour tracking, which ... more We propose and evaluate a novel method for enhancing performance of lips contour tracking, which is based on the concept of statistic shape models (ASM) and multi features. On the first image of the video sequence, lip region is detected using the Bayesian&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;s rule in which lip color information is modeled by using the Gaussian mixture model (GMM) and the

Adaptive boundary control of an axially moving string system: Application to container cranes

2009 IEEE International Symposium on Industrial Electronics, 2009

This paper proposes an adaptive boundary control to an axially moving string system, which couple... more This paper proposes an adaptive boundary control to an axially moving string system, which couples with a mass-damper-spring (MDS) controller at its right-hand-side (RHS) boundary. Unknown parameters appearing in the system equation are assumed constant and estimated on-line by using adaptation laws. The adaptive computed-torque control algorithm applied to robot manipulators of lumped systems is extended to design the adaptive boundary controller for the coupling system. It is found that the control force and update laws depend only on the displacement, velocity and slope of the string at the RHS boundary. Lyapunov stability guarantees the convergence of the tracking error to zero. Finally, the performance of the proposed controller is demonstrated by numerical simulations.

Models of tone for tonal and non-tonal languages

2013 IEEE Workshop on Automatic Speech Recognition and Understanding, 2013

Conventional wisdom in automatic speech recognition asserts that pitch information is not helpful... more Conventional wisdom in automatic speech recognition asserts that pitch information is not helpful in building speech recognizers for non-tonal languages and contributes only modestly to performance in speech recognizers for tonal languages. To maintain consistency between different systems, pitch is therefore often ignored, trading the slight performance benefits for greater system uniformity/ simplicity. In this paper, we report results that challenge this conventional approach. We present new models of tone that deliver consistent performance improvements for tonal languages (Cantonese, Vietnamese) and even modest improvements for non-tonal languages. Using neural networks for feature integration and fusion, these models achieve significant gains throughout, and provide us with system uniformity and standardization across all languages, tonal and non-tonal.

Multigrid Optimization Schemes for Solving Bose–Einstein Condensate Control Problems

SIAM Journal on Scientific Computing, 2008

The control of the transport of Bose-Einstein condensates in magnetic microtraps is formulated wi... more The control of the transport of Bose-Einstein condensates in magnetic microtraps is formulated within the framework of optimal control theory and solved by multigrid optimization schemes. The time evolution of the wave function of Bose-Einstein condensates is governed by the Gross-Pitaevskii equation and can be manipulated through variation of a controllable magnetic confinement potential. In order to define an optimal control strategy, an appropriate cost functional is introduced that must be minimized under the constraint given by the dynamic equation. The resulting optimality system consists of two nonlinear Schrödinger-type equations with opposite time orientation coupled with an elliptic equation for the control function. These equations are approximated by using a time-splitting pseudospectral method and finite differences. To solve the resulting problem a cascadic nonlinear conjugate gradient scheme and a multigrid optimization scheme are considered. The convergence properties of these two schemes are investigated theoretically, and their computational performance is discussed based on results of numerical experiments. It appears that the multigrid optimization scheme provides a robust optimization strategy.

An elliptic semilinear equation with source term involving boundary measures: the subcritical case

Revista Matemática Iberoamericana, 2000