Article
Statistics & Probability
Sheng Jiang, Surya T. Tokdar
Summary: Bayesian nonparametric regression with rescaled Gaussian process prior offers smoothness-adaptive function estimation with near minimax-optimal error rates. Hierarchical extensions equipped with stochastic variable selection also adapt to the unknown intrinsic dimension of a sparse true regression function, but it remains unclear if they offer variable selection consistency. Our result shows that variable consistency may be achieved in models where the true regression function has finite smoothness, leading to polynomially larger penalties for including false positive predictors.
ANNALS OF STATISTICS
(2021)
Article
Computer Science, Interdisciplinary Applications
Felipe Cabral Pinto, Johnathan G. Manchuk, Clayton V. Deutsch
Summary: Simulating spatial Gaussian realizations is essential in geostatistics and other fields involving uncertainty. In multivariate cases, different scales and spatial structures of geological variables complicate decorrelation methods. Blind source separation simplifies the multivariate problem by representing it as a linear combination of latent source variables with independent spatial structures.
COMPUTERS & GEOSCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Wei Huang, Richard Yi Da Xu
Summary: Context-aware recommender systems (CARS) leverage contextual information to generate more accurate recommendations, with the latent factors approach being dominant. Overfitting may be a challenge for standard GP model-based methods. To address this, a Gaussian Process Latent Variable Model Factorization (GPLVMF) method is proposed to improve accuracy and capture contextual importance.
PATTERN RECOGNITION LETTERS
(2021)
Article
Computer Science, Artificial Intelligence
Haitao Liu, Jiaqi Ding, Xinyu Xie, Xiaomo Jiang, Yusong Zhao, Xiaofang Wang
Summary: Multi-task regression leverages task similarity to improve prediction quality and reduce the need for big data. Gaussian process (GP) is used to create a non-parametric Bayesian multi-task regression paradigm. The linear model of coregionalization (LMC) is a well-known approach but has limitations. This study proposes a neural embedding of coregionalization and employs advanced variational inference and sparse approximation to enhance prediction quality and scalability.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Biochemical Research Methods
Jacob Williams, Shuangshuang Xu, Marco A. R. Ferreira
Summary: In this study, a novel Bayesian variable selection method based on nonlocal priors is proposed for genome-wide association studies. The method, called BGWAS, effectively reduces false positive rates while maintaining the ability to detect true positive SNPs. It achieves this through a two-step process of screening and model selection.
BMC BIOINFORMATICS
(2023)
Article
Plant Sciences
Xudong Zhao, Hanxu Wang, Hangyu Li, Yiming Wu, Guohua Wang
Summary: This study aims to select variables from a multidimensional feature to identify PPR proteins, proposing a variable selection framework and using a Gaussian mixture model for variable selection. Results indicate that certain variables play a crucial role in discriminating between PPR positive proteins and negative ones.
FRONTIERS IN PLANT SCIENCE
(2021)
Article
Mathematics
Chengxin Gong, Jinwen Ma
Summary: In this paper, an improved two-layer mixtures of Gaussian process functional regressions (TMGPFR) model is proposed, along with a Bayesian Ying-Yang (BYY) annealing learning algorithm for parameter learning and automated model selection. Experimental results show that the proposed algorithm can automatically make correct model selection during parameter learning.
Article
Biochemical Research Methods
Katrin Madjar, Manuela Zucknick, Katja Ickstadt, Joerg Rahnenfuehrer
Summary: The new Bayesian method provides separate risk prediction models for each cohort by sharing information between predictors to increase power. Results demonstrate that this approach outperforms standard methods in terms of prediction performance and power in variable selection especially when sample size is small.
BMC BIOINFORMATICS
(2021)
Article
Automation & Control Systems
Qingchao Jiang, Jiashi Jiang, Weimin Zhong, Xuefeng Yan
Summary: This study proposes a new Gaussian-process-based probabilistic latent variable (GPPLV) modeling framework for distributed monitoring of multiunit nonlinear processes. The proposed method performs better in showing the nature of different faults and has a higher fault detection rate for large-scale multiunit processes compared to some common distributed process monitoring models.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2023)
Article
Automation & Control Systems
Weiming Shao, Chuanfa Xiao, Jingbo Wang, Dongya Zhao, Zhihuan Song
Summary: This paper proposes a dynamic soft sensing method called 'semisupervised Bayesian hidden Markov model (SsBHMM)' to real-time sense key variables related to product quality in industrial processes. The SsBHMM utilizes a semisupervised fully Bayesian regressive model structure with a first-order Markov chain and mixture of Gaussians to account for process dynamics and non-Gaussian distributions. An efficient training algorithm based on variational inference is developed to learn parameters and handle missing values of quality variables. Numerical examples and an industrial low-temperature transformation unit demonstrate the advantages and feasibility of the SsBHMM.
JOURNAL OF PROCESS CONTROL
(2022)
Article
Construction & Building Technology
Seung-Seop Jin, Jinyoung Hong, Hajin Choi
Summary: The goal of this study is to improve sampling efficiency and achieve reasonable accuracy for the quantity of interest (damage) in impact-echo testing. The proposed solution is the implementation of Gaussian process (GP)-assisted active learning, which enables autonomous decision-making for data acquisition, identifies potential damage locations, and improves accuracy. Through numerical and experimental validation, this method demonstrates more accurate and efficient results.
AUTOMATION IN CONSTRUCTION
(2022)
Article
Automation & Control Systems
Hengrui Luo, Giovanni Nattino, Matthew T. Pratola
Summary: In this paper, a novel Gaussian process regression model is introduced in the fully Bayesian setting. The model integrates ideas of sparsification, localization, and Bayesian additive modeling, using a recursive partitioning scheme to fit sparse GP regression models and combining multiple layers of partitioned SGPs with efficient computations to capture global trends and local refinements.
JOURNAL OF MACHINE LEARNING RESEARCH
(2022)
Article
Geosciences, Multidisciplinary
Ying C. MacNab
Summary: This article reviews the development and current status of Bayesian disease mapping, discusses its importance in contemporary health science research, and explores the potential utility and impact of disease mapping models and methods for analyzing and monitoring communicable diseases.
SPATIAL STATISTICS
(2022)
Article
Mathematics
Junbo Zhang, Daoji Li, Yingzhi Xia, Qifeng Liao
Summary: This paper proposes a new two-step approach to estimate 1-km-resolution PM2.5 concentrations in Shanghai using satellite data. The approach refines AOD data to a higher resolution using a Bayesian retrieval method and then uses a hierarchical Gaussian process model to estimate PM2.5 concentrations. The results show accurate predictive performance of the proposed approach.
Article
Engineering, Mechanical
P. Gardner, T. J. Rogers, C. Lord, R. J. Barthorpe
Summary: This paper discusses the challenges and issues of predicting real world events with computer models, and proposes an approach for inferring model discrepancy to overcome bias in model calibration.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Environmental Sciences
Amruta Nori-Sarma, Rajesh K. Thimmulappa, G. V. Venkataramana, Azis K. Fauzie, Sumit K. Dey, Lalith K. Venkareddy, Jesse D. Berman, Kevin J. Lane, Kelvin C. Fong, Joshua L. Warren, Michelle L. Bell
ATMOSPHERIC ENVIRONMENT
(2020)
Article
Statistics & Probability
Joshua L. Warren, Thomas J. Luben, Howard H. Chang
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES C-APPLIED STATISTICS
(2020)
Article
Biology
Joshua L. Warren
JOURNAL OF AGRICULTURAL BIOLOGICAL AND ENVIRONMENTAL STATISTICS
(2020)
Article
Medicine, General & Internal
Kenneth S. Gunasekera, Jon Zelner, Mercedes C. Becerra, Carmen Contreras, Molly F. Franke, Leonid Lecca, Megan B. Murray, Joshua L. Warren, Ted Cohen
Article
Biotechnology & Applied Microbiology
Jordan Peccia, Alessandro Zulli, Doug E. Brackney, Nathan D. Grubaugh, Edward H. Kaplan, Arnau Casanovas-Massana, Albert I. Ko, Amyn A. Malik, Dennis Wang, Mike Wang, Joshua L. Warren, Daniel M. Weinberger, Wyatt Arnold, Saad B. Omer
NATURE BIOTECHNOLOGY
(2020)
Letter
Medicine, General & Internal
Anne L. Wyllie, John Fournier, Arnau Casanovas-Massana, Melissa Campbell, Maria Tokuyama, Pavithra Vijayakumar, Joshua L. Warren
NEW ENGLAND JOURNAL OF MEDICINE
(2020)
Article
Immunology
Maile T. Phillips, Joshua L. Warren, Noga Givon-Lavi, Adrienn Tothpal, Gili Regev-Yochay, Ron Dagan, Daniel M. Weinberger
Article
Statistics & Probability
Brian J. Reich, Yawen Guan, Denis Fourches, Joshua L. Warren, Stefanie E. Sarnat, Howard H. Chang
ANNALS OF APPLIED STATISTICS
(2020)
Article
Environmental Sciences
Qiong Wang, Huazhang Miao, Joshua L. Warren, Meng Ren, Tarik Benmarhnia, Luke D. Knibbs, Huanhuan Zhang, Qingguo Zhao, Cunrui Huang
Summary: Ozone exposure during the second trimester was found to be associated with an increased risk of term LBW in a retrospective cohort study conducted in Guangzhou. Monthly ozone exposure during the 4th-6th month (O-3-1 h) and the 6th month (O-3-8 h) was also correlated with LBW risk. Critical exposure windows were identified at the 15th-26th gestational weeks for O-3-1 h exposure and the 20th-26th weeks for O-3-8 h exposure, suggesting a potential vulnerability during these periods.
ENVIRONMENT INTERNATIONAL
(2021)
Article
Environmental Sciences
Nicole C. Deziel, Joshua L. Warren, Huang Huang, Haoran Zhou, Andreas Sjodin, Yawei Zhang
Summary: The study did not find a clear association between PCB or OCP exposure and PTC using single and multi-pollutant modeling. However, some associations were found in individuals born during peak production, suggesting further investigation into early-life exposures and subsequent thyroid cancer risk may be warranted.
ENVIRONMENTAL RESEARCH
(2021)
Article
Clinical Neurology
Richa Sharma, Lindsey R. Kuohn, Daniel M. Weinberger, Joshua L. Warren, Lauren H. Sansing, Adam Jasne, Guido Falcone, Amar Dhand, Kevin N. Sheth
Summary: Excess cerebrovascular deaths occurred during the COVID-19 pandemic in the United States, with decreased stroke-related EMS calls and increased time spent at home being associated factors. Public health measures are necessary to address the decrease in seeking medical care for acute stroke during the pandemic.
Article
Statistics & Probability
Joshua L. Warren, Marie Lynn Miranda, Joshua L. Tootoo, Claire E. Osgood, Michelle L. Bell
Summary: The study introduces spatial and spatiotemporal distributed lag data fusion methods for predicting ambient air pollution concentrations, incorporating predictive information from surrounding grid cells. Results show that the new methods often provide improved model fit and predictive accuracy when lagged information is beneficial. The code to apply these methods is available in the R package DLfuse.
ANNALS OF APPLIED STATISTICS
(2021)
Article
Public, Environmental & Occupational Health
Kayoko Shioda, Jiachen Cai, Joshua L. Warren, Daniel M. Weinberger
Summary: The study investigated the use of temporal and spatial aggregation methods to smooth control time series and adjust underlying trends, improving the ability of the synthetic control model to accurately assess vaccine impact.
Review
Public, Environmental & Occupational Health
Olivia Cords, Leonardo Martinez, Joshua L. Warren, Jamieson Michael O'Marr, Katharine S. Walter, Ted Cohen, Jimmy Zheng, Albert Ko, Julio Croda, Jason R. Andrews
Summary: Prisons are high-risk environments for tuberculosis, with significant disparities in the incidence and prevalence of tuberculosis in incarcerated populations across regions. People in prison are at elevated risk of contracting Mycobacterium tuberculosis infection and developing tuberculosis compared to the general population.
LANCET PUBLIC HEALTH
(2021)
Article
Ophthalmology
Jun Hui Lee, Anthony K. Ma, Joshua L. Warren, Christopher C. Teng
Summary: The study found that the increase in expenditure for iStent represented the majority of the growth in glaucoma surgical spending in the past decade, with higher male proportion and older age associated with faster adoption rates of iStent in states.
OPHTHALMOLOGY GLAUCOMA
(2021)