Search

Your search keyword '"Zhang, Lixiang"' showing total 633 results
Start Over Author "Zhang, Lixiang" Publication Type Periodicals
633 results on '"Zhang, Lixiang"'

2. Smoothing analysis of two-color distributive relaxation for solving 2D Stokes flow by multigrid method

Author

Zhu, Xingwen and Zhang, Lixiang

Subjects
Mathematics - Numerical Analysis
Abstract

Smoothing properties of two-color distributive relaxation for solving a two-dimensional (2D) Stokes flow by multigrid method are theoretically investigated by using the local Fourier analysis (LFA) method. The governing equation of the 2D Stokes flow in consideration is discretized with the non-staggered grid and an added pressure stabilization term with stabilized parameters to be determined is introduced into the discretization system in order to enhance the smoothing effectiveness in the analysis. So, an important problem caused by the added pressure stabilization term is how to determine a suitable zone of parameters in the added term. To that end, theoretically, a two-color distributive relaxation, developed on the two-color Jacobi point relaxation, is established for the 2D Stokes flow. Firstly, a mathematical constitution based on the Fourier modes with various frequency components is constructed as a base of the two-color smoothing analysis, in which the related Fourier representation is presented by the form of two-color Jacobi point relaxation. Then, an optimal one-stage relaxation parameter and related smoothing factor for the two-color distributive relaxation are applied to the discretization system, and an analytical expression of the parameter zone on the added pressure stabilization term is established by LFA. The obtained analytical results show that numerical schemes for solving 2D Stokes flow by multigrid method on the two-color distributive relaxation have a specific convergence zone on the parameters of the added pressure stabilization term, and the property of convergence is independent of mesh size, but depends on the parameters of the pressure stabilization term.

Published
2023

6. Robust DNN Surrogate Models with Uncertainty Quantification via Adversarial Training

Author

Zhang, Lixiang and Li, Jia

Subjects
Computer Science - Machine Learning
Abstract

For computational efficiency, surrogate models have been used to emulate mathematical simulators for physical or biological processes. High-speed simulation is crucial for conducting uncertainty quantification (UQ) when the simulation is repeated over many randomly sampled input points (aka, the Monte Carlo method). In some cases, UQ is only feasible with a surrogate model. Recently, Deep Neural Network (DNN) surrogate models have gained popularity for their hard-to-match emulation accuracy. However, it is well-known that DNN is prone to errors when input data are perturbed in particular ways, the very motivation for adversarial training. In the usage scenario of surrogate models, the concern is less of a deliberate attack but more of the high sensitivity of the DNN's accuracy to input directions, an issue largely ignored by researchers using emulation models. In this paper, we show the severity of this issue through empirical studies and hypothesis testing. Furthermore, we adopt methods in adversarial training to enhance the robustness of DNN surrogate models. Experiments demonstrate that our approaches significantly improve the robustness of the surrogate models without compromising emulation accuracy.

Published
2022

7. Identification of Key Node Sets in Tunneling Boring Machine Cutterhead Supply Chain Network Based on Deep Reinforcement Learning

Author

Li, Yinqian, Wen, Jingqian, Zhang, Yanzi, Zhang, Lixiang, Chaari, Fakher, Series Editor, Gherardini, Francesco, Series Editor, Ivanov, Vitalii, Series Editor, Haddar, Mohamed, Series Editor, Cavas-Martínez, Francisco, Editorial Board Member, di Mare, Francesca, Editorial Board Member, Kwon, Young W., Editorial Board Member, Tolio, Tullio A.M., Editorial Board Member, Trojanowska, Justyna, Editorial Board Member, Schmitt, Robert, Editorial Board Member, Xu, Jinyang, Editorial Board Member, Chien, Chen-Fu, editor, Dou, Runliang, editor, and Luo, Li, editor

Published
2024
Full Text
View/download PDF

8. Deep Reinforcement Learning for Dynamic Flexible Job-Shop Scheduling with Automated Guided Vehicles

Author

Hou, Zhenyu, Zhang, Lixiang, Wang, Yiheng, Hu, Yaoguang, Chaari, Fakher, Series Editor, Gherardini, Francesco, Series Editor, Ivanov, Vitalii, Series Editor, Haddar, Mohamed, Series Editor, Cavas-Martínez, Francisco, Editorial Board Member, di Mare, Francesca, Editorial Board Member, Kwon, Young W., Editorial Board Member, Tolio, Tullio A.M., Editorial Board Member, Trojanowska, Justyna, Editorial Board Member, Schmitt, Robert, Editorial Board Member, Xu, Jinyang, Editorial Board Member, Chien, Chen-Fu, editor, Dou, Runliang, editor, and Luo, Li, editor

Published
2024
Full Text
View/download PDF

21. Darboux transformations and solutions of nonlocal Hirota and Maxwell-Bloch equations

Author

An, Ling, Li, Chuanzhong, and Zhang, Lixiang

Subjects
Nonlinear Sciences - Exactly Solvable and Integrable Systems, Mathematical Physics, Physics - Optics, 37K05, 37K10, 35Q53
Abstract

In this paper, based on the Hirota and Maxwell-Bloch (H-MB) system and its application in the theory of the femtosecond pulse propagation through an erbium doped fiber, we define two kinds of nonlocal Hirota and Maxwell-Bloch (NH-MB) systems, namely, $PT$-symmetric NH-MB system and reverse space-time NH-MB system. Then we construct the Darboux transformations of these NH-MB systems. Meanwhile, we derive the explicit solutions by the Darboux transformations., Comment: 22 Pages, accepted for publication in Studies in Applied Mathematics

Published
2021

25. Deep Variable-Block Chain with Adaptive Variable Selection

Author

Zhang, Lixiang, Lin, Lin, and Li, Jia

Subjects
Statistics - Machine Learning, Computer Science - Machine Learning
Abstract

The architectures of deep neural networks (DNN) rely heavily on the underlying grid structure of variables, for instance, the lattice of pixels in an image. For general high dimensional data with variables not associated with a grid, the multi-layer perceptron and deep belief network are often used. However, it is frequently observed that those networks do not perform competitively and they are not helpful for identifying important variables. In this paper, we propose a framework that imposes on blocks of variables a chain structure obtained by step-wise greedy search so that the DNN architecture can leverage the constructed grid. We call this new neural network Deep Variable-Block Chain (DVC). Because the variable blocks are used for classification in a sequential manner, we further develop the capacity of selecting variables adaptively according to a number of regions trained by a decision tree. Our experiments show that DVC outperforms other generic DNNs and other strong classifiers. Moreover, DVC can achieve high accuracy at much reduced dimensionality and sometimes reveals drastically different sets of relevant variables for different regions., Comment: 24 pages, 5 figures

Published
2019

29. Tanshinone ⅡA attenuates high glucose-induced apoptosis and oxidative stress in mouse podocyte line PMC-5

Author

ZHANG Mengyao, NIU Shu, CAI Jing, ZHANG Lixiang, ZHAO Zhigang

Subjects
tanshinone ⅱa, podocytes, oxidative stress, apoptosis, high glucose, Medicine
Abstract

Objective To explore potential mechanisms involved in the reduction of podocyte apoptosis and oxidative stress induced by high glucose (HG) by tanshinone ⅡA (Tan ⅡA). Methods CCK-8 method was used to screen the appropriate interference concentration of Tan ⅡA. The MPC-5 cells were divided into NC group (5.5 mmol/L D-glucose), MA group(5.5 mmol/L D-glucose and 22.5 mmol/L mannitol), HG group(25 mmol/L D-glucose), HG+Tan ⅡA group(25 mmol/L D-glucose and 10 μmol/L Tan ⅡA), and HG+Tan ⅡA+SIRT1 inhibitors(sirtinol) group (25 mmol/L D-glucose, 10 μmol/L Tan ⅡA and 20 μmol/L sirtinol). Determina- tion of oxidative stress markers malondialdehyde(MDA), superoxide dismutase (SOD) and catalase (CAT) were made by microplate method, WST-1 method and visible light method. Flow cytometry was used to measure cell apoptosis. Western blot was used to evaluate the expression of apoptosis-related proteins and silent information regulator 1 (SIRT1)/endothelial nitric oxide synthase (eNOS) pathway proteins. Results Tan ⅡA improved the viability of MPC-5 cells induced by HG (P＜0.05). Compared with the MA group, the level of cell apoptosis in the HG group increased, the expression of B-cell lymphoma-2 gene (Bcl-2), phosphorylated SIRT1 (p-SIRT1) and phosphorylated eNOS(p-eNOS) in the cells, and the activity of SOD and of CAT were reduced, the expression of Bcl-2-associated X protein (Bax) and cleaved caspase-3, and the content of MDA in cells were increased(P＜0.05). Compared with the HG group, the level of cell apoptosis in the Tan ⅡA group reduced, the expressions of Bcl-2, p-SIRT1 and p-eNOS in the cells, and the activity of SOD and CAT were increased, the expression of Bax and cleaved caspase-3, and the content of MDA in cells were reduced(P＜0.05). The SIRT1 inhibitor sirtinol attenuated the inhibitory effect of Tan ⅡA on HG-induced podocyte apoptosis and oxidative stress (P＜0.05). Conclusions Tan ⅡA may reduce HG-induced apoptosis and oxidative stress in podocytes with a potential mechanism of activating the SIRT1/eNOS pathway.

Published
2023
Full Text
View/download PDF

37. Risk prediction model of in-hospital heart failure in patients with coronary heart disease after percutaneous coronary intervention: development and validation

Author

CAO Jiaoyu, ZHANG Lixiang, ZHOU Xiaojuan, YAO Xunxia, and MA Likun

Subjects
coronary heart disease, heart failure, risk prediction, nomogram, Medicine (General), R5-920
Abstract

Objective To analyze the risk factors for heart failure in hospitalized patients with coronary heart disease (CHD) after percutaneous coronary intervention (PCI) and establish a personalized risk prediction model for heart failure after PCI. Methods The clinical data of 640 patients with CHD who were hospitalized and underwent PCI in our department from January 2017 to December 2018 were collected and retrospectively analyzed. The subjects were divided into heart failure group (n=44) and non-heart failure group (n=596) according to the occurrence of postoperative heart failure. Logistic multivariate regression analysis and R software were used respectively to screen the independent predictors and plot the nomogram prediction model of heart failure risk. Then, consistency index (C-index), area under the receiver operating characteristic (ROC) curve and calibration curve were subsequently adopted to evaluate the prediction efficiency of the model. Results Multivariate analysis showed that left ventricular ejection fraction (LVEF), age, contrast agent dosage, concomitant cerebral infarction, and grade of New York Heart Association (NYHA) functional classification were independent predictors of heart failure after PCI (P < 0.05). The C-index of the nomogram for predicting heart failure risk was 0.796 (95%CI: 0.723~0.869), and the area under the ROC curve was 0.797 after internal validation (95%CI: 0.726~0.861). Hosmer-Lemeshow test and calibration curves indicated that the nomogram had good prediction consistency (Chi-square=12.360, P=0.135). Conclusion The nomogram model for predicting heart failure risk is successfully constructed, which can assist the medical staff to screen out the population with high risk of heart failure after PCI, and then to implement targeted intervention measures.

Published
2022
Full Text
View/download PDF

50. Predictive value of residual inflammation risk-based nomogram model for major adverse cardiovascular events in acute myocardial infarction patients after percutaneous coronary intervention

Author

CAO Jiaoyu, ZHANG Lixiang, ZHAN Ling, and MA Likun

Subjects
acute myocardial infarction, adverse cardiovascular events, residual inflammation, risk prediction, nomogram, Medicine (General), R5-920
Abstract

Objective To develop a nomogram prediction model based on residual inflammation risk (RIR) and evaluate its value in the prediction for the risk of major adverse cardiovascular events (MACE) in patients with acute myocardial infarction (AMI) after percutaneous coronary intervention (PCI). Methods Clinical data of 297 AMI patients who underwent PCI in our department from June 2017 to March 2019 were collected and retrospectively analyzed. According to their levels of preoperative hypersensitive-CRP (hs-CRP) and low-density lipoprotein cholesterol (LDL-C), they were divided into RIR group (n=28) and non RIR group (n=269). They also were assigned into the MACE group (n=102) and non-MACE group (n=195) according to whether MACEs occurred during hospitalization. Least absolute shrinkage and selection operator (LASSO) regression analysis and multivariate logistic regression analysis were used to evaluate the correlation between the risk of RIR and the occurrence of MACE, and other independent risk predictors of MACE were screened out. The nomogram model based on the indicators of RIR was constructed and verified by the R language software. Results The incidences of pump failure, cardiogenic shock, malignant arrhythmia and death, and overall MACE were significantly higher in the RIR group than the non-RIR group (P < 0.05). The Results of LASSO regression and multivariate regression analyses showed that left ventricular ejection fraction (LVEF) and hemoglobin concentration were negatively correlated with the risk of MACE in hospital (P < 0.05), while residual inflammation risk, hemoglobinA1c (Hb1Ac), leukocyte count and N-terminal pro-B type natriuretic peptide (NT-proBNP) level were positively correlated with the risk of MACE (P < 0.05). The ROC curve indicated that RIR was not good in prediction of MACE (AUC=0.592, 95%CI: 0.551~0.634). Based on the above 6 indicators, the nomogram model was constructed. The Harrel's C-index of the nomogram was 0.872 (95%CI: 0.827~0.917), the AUC of the nomogram after re-sampling 1 000 times was 0.866 (95% CI: 0.818~0.907), and the Hosmer lemeshow deviation test showed that the prediction probability of nomogram was consistent with the actual frequency (Chi-square=8.420, P=0.393). The clinical decision curve showed that when the threshold probability of MACE occurrence is between 0.08 and 0.88, and the nomogram model could obtain the highest net benefit, which indicated that the model had good clinical applicability. Conclusion The RIR-based nomogram model and other 5 factors have good prediction efficiency and clinical applicability in the prediction of the risk of MACE in AMI patients.

Published
2021
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results