A5016692079- Statistical Methods and Bayesian Inference
- Advanced Statistical Methods and Models
- Optimal Experimental Design Methods
- Statistical Methods in Clinical Trials
- Statistical Methods and Inference
- Bayesian Methods and Mixture Models
- Statistical Distribution Estimation and Applications
- Advanced Statistical Process Monitoring
- Water Quality and Resources Studies
- Bayesian Modeling and Causal Inference
- Gene expression and cancer classification
- Soil Geostatistics and Mapping
- Advanced Multi-Objective Optimization Algorithms
- Genetic and phenotypic traits in livestock
- Water Quality and Pollution Assessment
- Financial Risk and Volatility Modeling
- Network Security and Intrusion Detection
- Spam and Phishing Detection
- Forecasting Techniques and Applications
- Advanced Malware Detection Techniques
- Probabilistic and Robust Engineering Design
- Veterinary Orthopedics and Neurology
- Statistical Methods and Applications
- Non-Invasive Vital Sign Monitoring
- Bone health and osteoporosis research
Xiamen University
2025
The University of Texas at San Antonio
2011-2024
University of International Business and Economics
2022
The University of Texas Health Science Center at San Antonio
2014
Virginia Tech
1996-2005
Virginia–Maryland College of Veterinary Medicine
1999
University of Georgia
1999
Abstract Motivation: Large-scale gene expression profiling generates data sets that are rich in observed features but poor numbers of observations. The analysis such is a challenge has been object vigorous research. algorithms use for this purpose have poorly documented and rarely compared objectively, posing problem uncertainty about the outcomes analyses. One way to objectively test apply them on computational network models which mechanisms completely know. Results: We present system...
Estimating the number of clusters in a data set is crucial step cluster analysis. In this article, motivated by gap method (Tibshirani, Walther, and Hastie, 2001, Journal Royal Statistical Society B63, 411-423), we propose weighted difference difference-weighted (DD-weighted) methods for estimating using within-clusters sum errors: measure homogeneity. addition, "multilayer" clustering approach, which shown to be more accurate than original method, particularly detecting nested structure...
To assess water quality standards, measurements of under the Clean Water Act are collected on a regular basis over period time. The data analyzed to evaluate percentage samples exceeding standard. One problem is that current limited by time range and consequently sample size inadequate provide necessary precision in parameter estimation. address this issue, we present Bayesian approach using power prior incorporate historical and/or at adjacent stations. We develop modified discuss its...
Web threats pose the most significant cyber threat. Websites have been developed or manipulated by attackers for use as attack tools. Existing malicious website detection techniques can be classified into categories of static and dynamic approaches, which respectively aim to detect websites analyzing web contents, run-time behaviors using honeypots. However, existing approaches technical computational limitations sophisticated attacks analyze massive collected data. The main objective this...
Empirical models that relate multiple quality features to a set of design variables play vital role in many industrial process optimization methods. Many the current modeling methods employ single-response normal model analyze processes without taking into consideration high correlations and non-normality among response variables. Also, problem variable selection has also not yet been fully investigated within this framework. Failure account for these issues may result misleading prediction...
Enhancer clusters, pivotal in mammalian development and diseases, can organize as enhancer networks to control cell identity disease genes; however, the underlying mechanism remains largely unexplored. Here, we introduce eNet 2.0, a comprehensive tool for analysis during diseases based on single-cell chromatin accessibility data. 2.0 extends our previous work 1.0 by adding network topology, comparison dynamics analyses its construction function. We reveal modularly organized networks, where...
Malicious websites are a major cyber attack vector, and effective detection of them is an important defense task. The main paradigm in this regard that the defender uses some kind machine learning algorithms to train model, which then used classify question. Unlike other settings, following issue inherent problem malicious detection: attacker essentially has access same data his/her models. This `symmetry' can be exploited by attacker, at least principle, evade defender's In paper, we...
The authors examine family purchase-decision dynamics to shed light on enhancing marketing communication effectiveness. In particular, the are interested in understanding temporal nature of spousal behavioral interaction decision making help marketers target messages, shape brand choice, and guide personal selling activities. calibrate a dynamic simultaneous equations model investigate behavior: What interactions discrete purchase decision, what aspects behavior across decisions? results...
Section 303(d) of the Clean Water Act requires states to assess condition their waters and implement plans improve quality identified as impaired. U.S. Environmental Protection Agency guidelines require a stream segment be listed impaired when greater than 10% measurements water conditions exceed numeric criteria. This can termed "raw score" assessment approach. are samples taken from population conditions. Concentrations pollutants vary naturally, measurement errors may made, occasional...
Comparative calibration is the broad statistical methodology used to assess of a set p instruments, each designed measure same characteristic, on common group individuals. Different from usual problem, true underlying quantity measured unobservable. Many authors have shown that this in general, does not unique solution. Most commonly assumptions obtain solution are (i) one instrument gold standard (that is, unbiased) and (ii) measurement errors instruments independent. Such constraints,...
Environmental decision-making is complex and often based on multiple lines of evidence. Integrating the information from these evidence rarely a simple process. We present quantitative approach to combination through calculation weight-of-evidence, with reference conditions used define not impaired state. The risk-based measurement risk computed as probability impairment. When data are available, there variety methods for calculating this probability. Statistical theory use odds ratios...
(2005). Applied Bayesian Modeling and Causal Inference From Incomplete-Data Perspectives. Technometrics: Vol. 47, No. 4, pp. 519-519.
Abstract Objective To investigate effects of the use stance time or velocity as control variables on ground reaction forces in lame dogs. Animals 12 dogs with pelvic osteotomies. Procedure Data for were obtained preoperatively and at 2, 4, 6, 8, 12, 16, 20, 24, 28 weeks postoperatively, using variables. Ground compared between 2 methods data collection, velocities times trials. Results Significant differences not found a variable any time. Also, significant collection. Greatest variation was...
Berger & Bernardo's (1989) reference priors and Tibshirani are derived for distributions in Bar-Lev Reiser's (1982) two-parameter exponential family when either the location or scale parameter is of interest. The prior shown to be a prior. Furthermore, conditions under which higher-order matching investigated. When both parameters interest, it also that agrees with posterior frequentist expansions based on signed square root log-likelihood ratio. normal, inverse Gaussian, gamma used...
The power prior and its variations have been proven to be a useful class of informative priors in Bayesian inference due their flexibility incorporating the historical information by raising likelihood data fractional δ. derivation marginal based on original prior, variation, normalized introduces scaling factor C(δ) form predictive distribution with powered likelihood. In this article, we show that might infinite for some positive δ conventionally used initial priors, which would change...
Abstract Two reference priors for the product of means n normal distributions with common known variance are developed. One them induces an improper posterior distribution and therefore is not much interest. The other a generalized form = 2 case derived by Berger Bernardo. latter compared uniform prior (the Jeffreys prior) in inference optimal frequentist coverage criterion. shown to be better than sense correct probability quantile, numerical computation. computation was performed Gibbs...
In the exponential regression model, inference concerning parameter is notoriously difficult, even when using Bayesian noninformative prior approach. The reference approach (Bernardo, 1979; Berger & Bernardo, 1989) considered, and argued to yield very satisfactory inferences. Estimation credible sets are considered in a specific example.