Search

Your search keyword '"Shen, Changyu"' showing total 3,391 results
Start Over Author "Shen, Changyu"
3,391 results on '"Shen, Changyu"'

6. Bootstrapping the Cross-Validation Estimate

Author

Cai, Bryan, Pellegrini, Fabio, Pang, Menglan, de Moor, Carl, Shen, Changyu, Charu, Vivek, and Tian, Lu

Subjects
Statistics - Methodology, Mathematics - Statistics Theory, Statistics - Machine Learning
Abstract

Cross-validation is a widely used technique for evaluating the performance of prediction models. It helps avoid the optimism bias in error estimates, which can be significant for models built using complex statistical learning algorithms. However, since the cross-validation estimate is a random value dependent on observed data, it is essential to accurately quantify the uncertainty associated with the estimate. This is especially important when comparing the performance of two models using cross-validation, as one must determine whether differences in error estimates are a result of chance fluctuations. Although various methods have been developed for making inferences on cross-validation estimates, they often have many limitations, such as stringent model assumptions This paper proposes a fast bootstrap method that quickly estimates the standard error of the cross-validation estimate and produces valid confidence intervals for a population parameter measuring average model performance. Our method overcomes the computational challenge inherent in bootstrapping the cross-validation estimate by estimating the variance component within a random effects model. It is just as flexible as the cross-validation procedure itself. To showcase the effectiveness of our approach, we employ comprehensive simulations and real data analysis across three diverse applications.

Published
2023

7. Study of overflow effects on flow-induced stress in injection molding

Author

Lu, Zhili, Ma, Ruixue, Ma, Jialu, Cao, Wei, Huang, Qingwei, Zhang, Shixun, and Shen, Changyu

Subjects
Viscoelasticity -- Models, Injection molding -- Methods, Engineering and manufacturing industries, Science and technology
Abstract

Overflow is usually used to keep fluid continuously flowing to avoid the fluid from changing flow direction and subsequently inducing large stresses. To explore the mechanism of flow-induced stress evolution, a flow model was established in terms of compressible, non-isothermal, and viscoelastic flow. Based on the Leonov model and the mid-plane approach, a pseudo-Poisson type equation for pressure was derived to decouple the interdependence between velocity and pressure. To improve computational efficiency, a numerical approach involving double iterations was proposed to solve the flow problem. Two kinds of mold cavities, with and without overflow, were designed and manufactured. Molding experiments were conducted, and corresponding simulations based on the proposed model and algorithm were performed. The results show that the simulated contours of flow-induced stress are in good agreement with the photoelastic streaks. The use of overflow can significantly reduce flow-induced stresses. If the slot thickness is set to a proper value, the flow-induced stress can be reduced by more than 80%. On the other hand, a slot that is too thick does not have significant reducing effects. Therefore, using overflow to reduce flow-induced stresses during transparent part production is a promising approach if the material cost is worthy of the quality. Highlights * A double iteration method is proposed for viscoelastic flow problem. * Overflow can reduce the stress as much as 80% if the dimension is properly set. * The simulated stress streaks are in good agreement with photoelasticities. KEYWORDS finite element method, flow-induced stress, Leonov model, overflow, viscoelastic, 1 | INTRODUCTION Injection molding (IM) is the most efficient mass-production technology for manufacturing polymers in complex and detailed geometries with high precision. In view of these advantages, most transparent [...]

Published
2024
Full Text
View/download PDF

32. Comparative outcomes for mature T-cell and NK/T-cell lymphomas in people with and without HIV and to AIDS-defining lymphomas

Author

Koh, Min Jung, Merrill, Mwanasha H, Koh, Min Ji, Stuver, Robert, Alonso, Carolyn D, Foss, Francine M, Mayor, Angel M, Gill, John, Epeldegui, Marta, Cachay, Edward, Thorne, Jennifer E, Silverberg, Michael J, Horberg, Michael A, Althoff, Keri N, Nijhawan, Ank E, McGinnis, Kathleen A, Lee, Jennifer S, Rabkin, Charles S, Napravnik, Sonia, Li, Jun, Castilho, Jessica L, Shen, Changyu, and Jain, Salvia

Subjects
Biomedical and Clinical Sciences, Cardiovascular Medicine and Haematology, Oncology and Carcinogenesis, Lymphoma, Sexually Transmitted Infections, Lymphatic Research, Infectious Diseases, Hematology, Rare Diseases, HIV/AIDS, Cancer, Acquired Immunodeficiency Syndrome, Hodgkin Disease, Humans, Lymphoma, AIDS-Related, Lymphoma, Large B-Cell, Diffuse, Lymphoma, T-Cell, Peripheral, T-Lymphocytes, Cardiovascular medicine and haematology
Abstract

There are no studies comparing the prognosis for mature T-cell lymphoma (TCL) in people with HIV (PWH) to people without HIV (PWoH) and to AIDS-defining B-cell lymphomas (A-BCLs) in the modern antiretroviral therapy era. North American AIDS Cohort Collaboration on Research and Design and Comprehensive Oncology Measures for Peripheral T-cell Lymphoma Treatment are cohorts that enroll patients diagnosed with HIV and TCL, respectively. In our study, 52, 64, 101, 500, and 246 PWH with histologic confirmation of TCL, primary central nervous system lymphoma, Burkitt's lymphoma, diffuse large B-cell lymphoma (DLBCL), and Hodgkin's lymphoma (HL), respectively, and 450 TCLs without HIV were eligible for analysis. At the time of TCL diagnosis, anaplastic large-cell lymphoma (ALCL) was the most common TCL subtype within PWH. Although PWH with TCL diagnosed between 1996 and 2009 experienced a low 5-year survival probability at 0.23 (95% confidence interval [CI]: 0.13, 0.41), we observed a marked improvement in their survival when diagnosed between 2010 and 2016 (0.69; 95% CI: 0.48, 1; P = .04) in contrast to TCLs among PWoH (0.45; 95% CI: 0.41, 0.51; P = .53). Similarly, PWH with ALCLs diagnosed between 1996 and 2009 were associated with a conspicuously inferior 5-year survival probability (0.17; 95% CI: 0.07, 0.42) and consistently lagged behind A-BCL subtypes such as Burkitt's (0.43; 95% CI:0.33, 0.57; P = .09) and DLBCL (0.17; 95% CI: 0.06, 0.46; P = .11) and behind HL (0.57; 95% CI: 0.50, 0.65; P < .0001). Despite a small number, those diagnosed between 2010 and 2016 experienced a remarkable improvement in survival (0.67; 95% CI: 0.3, 1) in comparison with PWoH (0.76; 95% CI: 0.66, 0.87; P = .58). Thus, our analysis confirms improved overall survival for aggressive B- and T-cell malignancies among PWH in the last decade.

Published
2022

39. Improvement of image classification by multiple optical scattering

Author

Gao, Xinyu, Li, Yi, Qiu, Yanqing, Mao, Bangning, Chen, Miaogen, Meng, Yanlong, Zhao, Chunliu, Kang, Juan, Guo, Yong, and Shen, Changyu

Subjects
Computer Science - Computer Vision and Pattern Recognition, Electrical Engineering and Systems Science - Image and Video Processing
Abstract

Multiple optical scattering occurs when light propagates in a non-uniform medium. During the multiple scattering, images were distorted and the spatial information they carried became scrambled. However, the image information is not lost but presents in the form of speckle patterns (SPs). In this study, we built up an optical random scattering system based on an LCD and an RGB laser source. We found that the image classification can be improved by the help of random scattering which is considered as a feedforward neural network to extracts features from image. Along with the ridge classification deployed on computer, we achieved excellent classification accuracy higher than 94%, for a variety of data sets covering medical, agricultural, environmental protection and other fields. In addition, the proposed optical scattering system has the advantages of high speed, low power consumption, and miniaturization, which is suitable for deploying in edge computing applications.

Published
2021

Catalog

Books, media, physical & digital resources
See catalog results