A5053171184- Advanced Multi-Objective Optimization Algorithms
- Probabilistic and Robust Engineering Design
- Optimal Experimental Design Methods
- Simulation Techniques and Applications
- Advanced Statistical Methods and Models
- Gaussian Processes and Bayesian Inference
- Infrastructure Maintenance and Monitoring
- Bayesian Modeling and Causal Inference
- Evolutionary Algorithms and Applications
- Machine Learning and Algorithms
- Statistical Methods and Inference
- Statistical Methods and Bayesian Inference
- Health Systems, Economic Evaluations, Quality of Life
- Fuzzy Systems and Optimization
- Fault Detection and Control Systems
- Traffic Prediction and Management Techniques
- Environmental Impact and Sustainability
- Advanced Statistical Process Monitoring
- Risk and Safety Analysis
- Model Reduction and Neural Networks
- Traffic control and management
- Healthcare Policy and Management
- Bayesian Methods and Mixture Models
- Structural Health Monitoring Techniques
- Scientific Research and Discoveries
University of Lisbon
2007-2024
HES-SO University of Applied Sciences and Arts Western Switzerland
2022
eBOS Technologies (Cyprus)
2022
Ionian University
2022
National and Kapodistrian University of Athens
2022
Schott (Germany)
2022
University of Kaiserslautern
2022
Institute of Communication and Computer Systems
2022
National Technical University of Athens
2022
Chemnitz University of Technology
2022
AbstractZellner's g prior remains a popular conventional for use in Bayesian variable selection, despite several undesirable consistency issues. In this article we study mixtures of priors as an alternative to default that resolve many the problems with original formulation while maintaining computational tractability has made so popular. We present theoretical properties mixture and provide real simulated examples compare fixed priors, empirical Bayes approaches, other procedures. Please...
AbstractWe present a framework that enables computer model evaluation oriented toward answering the question: Does adequately represent reality? The proposed validation is six-step procedure based on Bayesian and likelihood methodology. methodology particularly well suited to treating major issues associated with process: quantifying multiple sources of error uncertainty in models, combining information, updating assessments as new information acquired. Moreover, it allows inferential...
A key question in evaluation of computer models is Does the model adequately represent reality? six-step process for validation set out Bayarri et al. [Technometrics 49 (2007) 138–154] (and briefly summarized below), based on comparison runs with field data being modeled. The methodology particularly suited to treating major issues associated process: quantifying multiple sources error and uncertainty models; combining information; able adapt different, but related scenarios. Two...
Motivated by the statistical evaluation of complex computer models, we deal with issue objective prior specification for parameters Gaussian processes. In particular, derive Jeffreys-rule, independence Jeffreys and reference priors this situation, prove that resulting posterior distributions are proper under a quite general set conditions. A flat strategy, based on maximum likelihood estimates, is also considered, all then compared grounds frequentist properties ensuing Bayesian procedures....
Two different approaches to the prediction problem are compared employing a realistic example---combustion of natural gas---with 102 uncertain parameters and 76 quantities interests. One approach, termed bound-to-bound data collaboration (abbreviated B2B), deploys semidefinite programming algorithms where initial bounds on unknowns combined with experimental produce new uncertainty for that consistent and, finally, deterministic in settings. The other approach is statistical Bayesian,...
Abstract The CRASH computer model simulates the effect of a vehicle colliding against different barrier types. If it accurately represents real crashworthiness, can be great value in various aspects design, such as setting timing air bag releases. goal this study is to address problem validating for design goals, based on utilizing runs and experimental data from crashes. This task complicated by fact that (i) output consists smooth functional data, (ii) certain types collision have very...
This paper introduces the R package SAVE which implements statistical methodology for analysis of computer models. Namely, includes routines that perform emulation, calibration and validation this type The is Bayesian essentially Bayarri, Berger, Paulo, Sacks, Cafeo, Cavendish, Lin, Tu (2007). available through Comprehensive Archive Network. We illustrate its use with a real data example in context simulated example.
In the context of a Gaussian multiple regression model, we address problem variable selection when in list potential predictors there are factors, that is, categorical variables. We adopt model perspective, approach by constructing class models, each corresponding to particular active The methodology is Bayesian and proceeds computing posterior probability these models. highlight fact set competing models depends on dummy representation an issue already documented Fernández et al. example...
Traditionally, screening refers to the problem of detecting influential (active) inputs in computer model. We develop methodology that applies screening, but main focus is on active not model itself rather discrepancy function introduced account for inadequacy when linking with field observations. contend this an important as it informs modeler which are potentially being mishandled model, also along directions may be less recommendable use prediction. The Bayesian and inspired by continuous...
Calibration and validation of traffic models are processes that depend on field data often limited but essential for determination inputs to the model assessment its reliability. Quantification systematization calibration process expose statistical issues inherent in use such data. Formalization naturally leads Bayesian methodology uncertainties predictions arise from a multiplicity sources, especially variability estimation input parameters discrepancy. The general problem was elucidated an...
Calibration and validation of traffic models are processes that depend on field data often limited but essential for determination inputs to the model assessment its reliability. Quantification systematization calibration process expose statistical issues inherent in use such data. Formalization naturally leads Bayesian methodology uncertainties predictions arise from a multiplicity sources, especially variability estimation input parameters discrepancy. The general problem was elucidated an...
In the construction of input–output models from supply-use tables, technology assumptions disambiguate how an industry uses inputs in production recipe multiple outputs. This paper Bayes' theorem to select assumptions, taking into account empirical observations. The presents a formulation explore hybrids between product and product-by-product tables. We then present Markov chain Monte-Carlo techniques implement Bayesian method for selecting assumptions. apply case study using Eurostat tables...
Factors are categorical variables, and the values which these variables assume called levels. In this paper, we consider variable selection problem where set of potential predictors contains both factors numerical variables. Formally, is a particular case standard coded using dummy As such, Bayesian solution would be straightforward and, possibly because this, problem, despite its importance, has not received much attention in literature. Nevertheless, show that perception illusory fact...
The changes that the Affordable Care Act introduced to US health insurance market have entirely altered traditional ratemaking process. Precisely, creation of statewide community rating schemes and a guaranteed issue has facilitated coverage high-risk population, leading massive in risk pool compositions. implementation Risk Adjustment neutralized some consequences limiting premium variation market. However, setting appropriate rate levels remained cumbersome due uncertainty about pool. Many...
Screening traditionally refers to the problem of detecting active inputs in computer model. In this paper, we develop methodology that applies screening, but main focus is on not model itself rather discrepancy function introduced account for inadequacy when linking with field observations. We contend an important as it informs modeler which are potentially being mishandled model, also along directions may be less recommendable use prediction. The Bayesian and inspired by continuous spike...