Abstract:Uncertainty is ubiquitous in nature, engineering systems, and our society. It may arise from incomplete representations of the underlying processes, unknown parameters and inputs to the mathematical models, approximations of numerical schemes, and many other factors. Recent advancements in mathematical modeling and data and computational sciences enable one to address those issues through the study of Uncertainty Quantification (UQ), a developing field “bringing to fruition the dream of being able to accurately model and predict real complex processes through computational simulators” [1]. Inherently interdisciplinary, an effective UQ requires not only the domain knowledge of specific processes and data to construct mathematical models, its implementation also seeks a host of mathematical and statistical tools with high-performance computation. To meet the growing interests of UQ from various industrial and research communities, the Society for Industrial and Applied Mathematics started the first biannual meeting in 2014, the SIAM Conference on Uncertainty Quantification. In 2015, a new conference called Quantification of Uncertainty in Engineering, Sciences and Technology (QUEST) was initiated in China. The goal of the meeting is to bring together the leading researchers from a large variety of disciplines to address recent development and challenges in many practical fields. Over the three-day conference from October 19 to 23, 2015, keynote speeches were given by the top UQ experts such as Dr. Roger Ghanem, Dr. George Karniadakis, Dr. Nicholas Zabaras, Dr. Wei Chen, Dr. Dongbin Xiu, and Dr. George Christakos. It also attracted 40 presentations ranging from environmental sciences, epidemiology, energy systems, and biomedical engineering to social sciences. We would like to thank theInternational Journal for Uncertainty Quantification , the first professional journal dedicated to UQ (its impact factor a year after the journal started being evaluated doubled to 1.0), for giving us the opportunity to prepare this special issue in conjunction with the QUEST Conference in 2015. The key focus of the present special issue is to highlight the demand of UQ in many areas and includes seven peer-reviewed articles from a multidisciplinary perspective. The first paper by Huo and coauthors examines the explosive critical phenomena of combinational phases on multiplex networks. By applying the Kuramoto model on multiplex networks, they found an abrupt transition in the synchronization of combinational phases and presents the phases of oscillators. The second paper by Wang, Wu, and Xiao investigates model-form uncertainty reduction in Reynolds-averaged Navier-Stokes (RANS) models in turbulent flows. To be specific, they propose a Bayesian framework by incorporating various types of prior knowledge and show that informative physics-based prior plays an important role in model prediction when the observation data are limited. Zhang, Duan, and Peng propose an optimization scheme to match scenes in navigation systems. By first identifying the matching interval and points, they design a cost function in accordance with required accuracy and determine the filtering parameters using local sensitivity analysis. Their work provides a theoretical basis for engineering applications. The paper by Xu and coauthors studies the software reliability growth model in a network environment. Treating the collective environmental uncertainty as temporally correlated noise, they provide a semianalytical solution of fault removal number in the software fault detection process, which is often modeled as the nonhomogeneous Poisson process (NHPP). Experimental comparisons with existing methods demonstrate that the new framework shows a closer fit to actual data and exhibits a more accurately predictive power. Zhao, Yang, and Xiao introduce a repeated-kernel-density-estimation-based approach for constructing evidence bodies from uncertain observations and interval measurements. Using kernel density estimation with a loop, they obtain a family of probability distribution based on given observations and discretize the probability box, characterized by the bounds of the probability distribution family, using an outer discretization method. The paper by Shi and coauthors focuses on the peer review process of scientific proposal evaluations in China, which may be subject to uncertainty including individual reviewers bias and potential discussion under the table. By

New Metrics for Validation of Data-Driven Random Process Models in Uncertainty Quantification

A New Validation Metric for Models with Correlated Responses Using Limited Experimental and Simulation Data

Comparative Study of Uncertainty Quantification Metrics via a Stochastic Method of Model Validation

Model Validation Via Uncertainty Propagation and Data Transformations

A Validation Metric for Model with Mixture of Random and Interval Variables

Toward a Better Understanding of Model Validation Metrics

Validation metric for multi-output model with mixed uncertainty of random and interval variables

Validation metric for single output models with stochastic and interval mixed uncertainty

Quantitative model validation techniques: new insights

Quality measures for the evaluation of machine learning architectures on the quantification of epistemic and aleatoric uncertainties in complex dynamical systems

Validation of uncertainty quantification metrics: a primer based on the consistency and adaptivity concepts

New validation metrics for models with multiple correlated responses

Validation Metric Based on Mahalanobis Distance for Models with Multiple Correlated Responses.

An Optimal Model Using Data Envelopment Analysis for Uncertainty Metrics in Reliability.

Evaluation of machine learning architectures on the quantification of epistemic and aleatoric uncertainties in complex dynamical systems

Quantifying Model Uncertainties in the Space of Probability Measures

Bayesian Validation Assessment of Multivariate Computational Models

PREFACE: FIRST QUEST CONFERENCE

A Probabilistic Quality-Relevant Monitoring Method with Gaussian Mixture Model

An Optimal Evaluating Method for Uncertainty Metrics in Reliability Based on Uncertain Data Envelopment Analysis.