arXiv subjects

Samuel W. K. Wong

Publications and source records attributed to Samuel W. K. Wong.

At least 19 recordsLinked to original sources

arXiv2026

A Bayesian Weakest-Link Framework for Joint Estimation of Material Strength and Stress Profile

For structural components whose failure is governed by the weakest-link theory, existing reliability models typically either assume that the underlying mechanical model is known or neglect to exploit the spatial information contained in observed failure locations. In practice, however, idealized mechanical models may systematically deviate from the actual stress due to simplifying or incorrect assumptions. To address this limitation, we propose a hierarchical Bayesian weakest-link model that jointly estimates the latent material strength and stress profile from paired failure load and failure zone observations. In our formulation, the stress profile is estimated via a B-spline basis expansion, and the non-differentiable weakest-link mechanism is approximated by a differentiable Softmin function to account for unobserved material flaws and facilitate Bayesian inference. Simulation studies demonstrate that the proposed framework provides accurate and robust estimation under various experiment configurations. Applied to a real-data analysis of Douglas-fir crossarms, the proposed model identifies systematic deviations from idealized beam theory that cannot be captured by deterministic stress derivations.

stat.ME↗

arXiv2026

Fully Bayesian Sequential Design for Heteroscedastic Stochastic Simulations

We present a fully Bayesian sequential strategy for predicting the mean response surface of heteroscedastic stochastic simulation models. Leveraging dual Gaussian processes as the surrogate and a selection criterion based on expected Bayesian integrated mean-square prediction error, our approach sequentially selects informative design points while fully accounting for parameter uncertainty. Sequential importance sampling is employed to efficiently update the posterior distribution of the parameters. Our strategy is tailored for expensive simulation models, where achieving robust predictive accuracy under a limited budget is critical. Using synthetic examples, we illustrate its practical advantages compared to existing approaches, in terms of predictive accuracy, noise estimation, and uncertainty quantification. We then implement the proposed strategy on a real motivating application in seismic design of wood-frame podium buildings.

stat.ME↗

arXiv2025

Understanding the Impact of Seasonal Climate Change on Canada's Economy by Region and by Sector

To assess the impact of climate change on the Canadian economy, we investigate the relationship between seasonal climate variables and economic growth across provinces and economic sectors. We also provide projections of climate change impacts up to the year of 2050, taking into account the diverse climate change patterns and economic conditions across Canada. Our results indicate that rising Winter temperature anomalies have a notable adverse impact on Canadian economic growth. Province-wide, Quebec, Manitoba, and Ontario are anticipated to experience larger negative impacts, whereas British Columbia is less vulnerable. Industry-wide, Finance and Real Estate, Science and Technology, and Information, Culture and Recreation are consistently projected to see mild benefits, while adverse effects are predicted for Manufacturing, Agriculture, and Mining. The disparities of climate change effects between provinces and industries highlight the need for governments to tailor their policies accordingly, and offer targeted assistance to regions and industries that are particularly vulnerable in the face of climate change. Targeted approaches to climate change mitigation are likely to be more effective than one-size-fits-all policies for the whole economy.

stat.AP↗

arXiv2025

FECT: Factuality Evaluation of Interpretive AI-Generated Claims in Contact Center Conversation Transcripts

Large language models (LLMs) are known to hallucinate, producing natural language outputs that are not grounded in the input, reference materials, or real-world knowledge. In enterprise applications where AI features support business decisions, such hallucinations can be particularly detrimental. LLMs that analyze and summarize contact center conversations introduce a unique set of challenges for factuality evaluation, because ground-truth labels often do not exist for analytical interpretations about sentiments captured in the conversation and root causes of the business problems. To remedy this, we first introduce a \textbf{3D} -- \textbf{Decompose, Decouple, Detach} -- paradigm in the human annotation guideline and the LLM-judges' prompt to ground the factuality labels in linguistically-informed evaluation criteria. We then introduce \textbf{FECT}, a novel benchmark dataset for \textbf{F}actuality \textbf{E}valuation of Interpretive AI-Generated \textbf{C}laims in Contact Center Conversation \textbf{T}ranscripts, labeled under our 3D paradigm. Lastly, we report our findings from aligning LLM-judges on the 3D paradigm. Overall, our findings contribute a new approach for automatically evaluating the factuality of outputs generated by an AI system for analyzing contact center conversations.

cs.CL↗

arXiv2025

Manifold-Constrained Gaussian Processes for Inference of Mixed-effects Ordinary Differential Equations with Application to Pharmacokinetics

Pharmacokinetic modeling using ordinary differential equations (ODEs) has an important role in dose optimization studies, where dosing must balance sustained therapeutic efficacy with the risk of adverse side effects. Such ODE models characterize drug plasma concentration over time and allow pharmacokinetic parameters to be inferred, such as drug absorption and elimination rates. For time-course studies involving treatment groups with multiple subjects, mixed-effects ODE models are commonly used. However, existing methods tend to lack uncertainty quantification on a subject-level, for key measures such as peak or trough concentration and for making predictions of drug concentration. To address such limitations, we propose an extension of manifold-constrained Gaussian processes for inference of general mixed-effects ODE models within a Bayesian statistical framework. We evaluate our method on simulated examples, demonstrating its ability to provide fast and accurate inference for parameters and trajectories using nested optimization. To illustrate the practical efficacy of the proposed method, we provide a real data analysis of a pharmacokinetic model used for an HIV combination therapy study.

stat.ME↗

arXiv2024

Inference for Delay Differential Equations Using Manifold-Constrained Gaussian Processes

Dynamic systems described by differential equations often involve feedback among system components. When there are time delays for components to sense and respond to feedback, delay differential equation (DDE) models are commonly used. This paper considers the problem of inferring unknown system parameters, including the time delays, from noisy and sparse experimental data observed from the system. We propose an extension of manifold-constrained Gaussian processes to conduct parameter inference for DDEs, whereas the time delay parameters have posed a challenge for existing methods that bypass numerical solvers. Our method uses a Bayesian framework to impose a Gaussian process model over the system trajectory, conditioned on the manifold constraint that satisfies the DDEs. For efficient computation, a linear interpolation scheme is developed to approximate the values of the time-delayed system outputs, along with corresponding theoretical error bounds on the approximated derivatives. Two simulation examples, based on Hutchinson's equation and the lac operon system, together with a real-world application using Ontario COVID-19 data, are used to illustrate the efficacy of our method.

stat.ME↗

arXiv2024

Spatio-temporal data fusion for the analysis of in situ and remote sensing data using the INLA-SPDE approach

We propose a Bayesian hierarchical model to address the challenge of spatial misalignment in spatio-temporal data obtained from in situ and satellite sources. The model is fit using the INLA-SPDE approach, which provides efficient computation. Our methodology combines the different data sources in a "fusion"" model via the construction of projection matrices in both spatial and temporal domains. Through simulation studies, we demonstrate that the fusion model has superior performance in prediction accuracy across space and time compared to standalone "in situ" and "satellite" models based on only in situ or satellite data, respectively. The fusion model also generally outperforms the standalone models in terms of parameter inference. Such a modeling approach is motivated by environmental problems, and our specific focus is on the analysis and prediction of harmful algae bloom (HAB) events, where the convention is to conduct separate analyses based on either in situ samples or satellite images. A real data analysis shows that the proposed model is a necessary step towards a unified characterization of bloom dynamics and identifying the key drivers of HAB events.

stat.ME↗

arXiv2023

Particle Gibbs for Likelihood-Free Inference of State Space Models with Application to Stochastic Volatility

State space models (SSMs) are widely used to describe dynamic systems. However, when the likelihood of the observations is intractable, parameter inference for SSMs cannot be easily carried out using standard Markov chain Monte Carlo or sequential Monte Carlo methods. In this paper, we propose a particle Gibbs sampler as a general strategy to handle SSMs with intractable likelihoods in the approximate Bayesian computation (ABC) setting. The proposed sampler incorporates a conditional auxiliary particle filter, which can help mitigate the weight degeneracy often encountered in ABC. To illustrate the methodology, we focus on a classic stochastic volatility model (SVM) used in finance and econometrics for analyzing and interpreting volatility. Simulation studies demonstrate the accuracy of our sampler for SVM parameter inference, compared to existing particle Gibbs samplers based on the conditional bootstrap filter. As a real data application, we apply the proposed sampler for fitting an SVM to S&P 500 Index time-series data during the 2008 financial crisis.

stat.ME↗

arXiv2023

A Bayesian Collocation Integral Method for Parameter Estimation in Ordinary Differential Equations

Inferring the parameters of ordinary differential equations (ODEs) from noisy observations is an important problem in many scientific fields. Currently, most parameter estimation methods that bypass numerical integration tend to rely on basis functions or Gaussian processes to approximate the ODE solution and its derivatives. Due to the sensitivity of the ODE solution to its derivatives, these methods can be hindered by estimation error, especially when only sparse time-course observations are available. We present a Bayesian collocation framework that operates on the integrated form of the ODEs and also avoids the expensive use of numerical solvers. Our methodology has the capability to handle general nonlinear ODE systems. We demonstrate the accuracy of the proposed method through simulation studies, where the estimated parameters and recovered system trajectories are compared with other recent methods. A real data example is also provided.

stat.ME↗

arXiv2023

MAGI: A Package for Inference of Dynamic Systems from Noisy and Sparse Data via Manifold-constrained Gaussian Processes

This article presents the MAGI software package for the inference of dynamic systems. The focus of MAGI is on dynamics modeled by nonlinear ordinary differential equations with unknown parameters. While such models are widely used in science and engineering, the available experimental data for parameter estimation may be noisy and sparse. Furthermore, some system components may be entirely unobserved. MAGI solves this inference problem with the help of manifold-constrained Gaussian processes within a Bayesian statistical framework, whereas unobserved components have posed a significant challenge for existing software. We use several realistic examples to illustrate the functionality of MAGI. The user may choose to use the package in any of the R, MATLAB, and Python environments.

stat.CO↗

arXiv2023

Knots and their effect on the tensile strength of lumber: a case study

When assessing the strength of sawn lumber for use in engineering applications, the sizes and locations of knots are an important consideration. Knots are the most common visual characteristics of lumber, that result from the growth of tree branches. Large individual knots, as well as clusters of distinct knots, are known to have strength-reducing effects. However, industry grading rules that govern knots are informed by subjective judgment to some extent, particularly the spatial interaction of knots and their relationship with lumber strength. This case study reports the results of an experiment that investigated and modelled the strength-reducing effects of knots on a sample of Douglas Fir lumber. Experimental data were obtained by taking scans of lumber surfaces and applying tensile strength testing. The modelling approach presented incorporates all relevant knot information in a Bayesian framework, thereby contributing a more refined way of managing the quality of manufactured lumber.

stat.AP↗

arXiv2023

Estimating and Assessing Differential Equation Models with Time-Course Data

Ordinary differential equation (ODE) models are widely used to describe chemical or biological processes. This article considers the estimation and assessment of such models on the basis of time-course data. Due to experimental limitations, time-course data are often noisy and some components of the system may not be observed. Furthermore, the computational demands of numerical integration have hindered the widespread adoption of time-course analysis using ODEs. To address these challenges, we explore the efficacy of the recently developed MAGI (MAnifold-constrained Gaussian process Inference) method for ODE inference. First, via a range of examples we show that MAGI is capable of inferring the parameters and system trajectories, including unobserved components, with appropriate uncertainty quantification. Second, we illustrate how MAGI can be used to assess and select different ODE models with time-course data based on MAGI's efficient computation of model predictions. Overall, we believe MAGI is a useful method for the analysis of time-course data in the context of ODE models, which bypasses the need for any numerical integration.

q-bio.MN↗

arXiv2023

A Kriging Metamodel with Adaptive Sampling for Seismic Evaluation of Podium Buildings

In this paper, nonlinear time-history dynamic analyses of selected earthquake ground motions are conducted on designated wood-frame podium buildings and the resulting inter-story drifts are analyzed. We aim to construct a reliable region where performance-based seismic design criteria are met, such that a two-step analysis procedure can be used with high confidence. We develop a kriging metamodel with tailored adaptive sampling methods to achieve this goal in a computationally efficient manner. The input variables we consider are the normalized stiffness ratio and the normalized mass ratio of the podium building. We took a six-story wood frame built upon a one-story concrete podium as a case study for our methodology, where our results indicate that the two-step analysis procedure may be used with high confidence if its normalized stiffness ratio is at least 38 and its normalized mass ratio is between 0.5 and 1.5.

stat.AP↗

arXiv2022

Estimating Boltzmann Averages for Protein Structural Quantities Using Sequential Monte Carlo

Sequential Monte Carlo (SMC) methods are widely used to draw samples from intractable target distributions. Particle degeneracy can hinder the use of SMC when the target distribution is highly constrained or multimodal. As a motivating application, we consider the problem of sampling protein structures from the Boltzmann distribution. This paper proposes a general SMC method that propagates multiple descendants for each particle, followed by resampling to maintain the desired number of particles. Simulation studies demonstrate the efficacy of the method for tackling the protein sampling problem. As a real data example, we use our method to estimate the number of atomic contacts for a key segment of the SARS-CoV-2 viral spike protein.

stat.ME↗

arXiv2022

A Comparative Study of Compartmental Models for COVID-19 Transmission in Ontario, Canada

The number of confirmed COVID-19 cases reached over 1.3 million in Ontario, Canada by June 4, 2022. The continued spread of the virus underlying COVID-19 has been spurred by the emergence of variants since the initial outbreak in December, 2019. Much attention has thus been devoted to tracking and modelling the transmission of COVID-19. Compartmental models are commonly used to mimic epidemic transmission mechanisms and are easy to understand. Their performance in real-world settings, however, needs to be more thoroughly assessed. In this comparative study, we examine five compartmental models -- four existing ones and an extended model that we propose -- and analyze their ability to describe COVID-19 transmission in Ontario from January 2022 to June 2022.

q-bio.PE↗

arXiv2022

Monte Carlo sampling of flexible protein structures: an application to the SARS-CoV-2 omicron variant

Proteins can exhibit dynamic structural flexibility as they carry out their functions, especially in binding regions that interact with other molecules. For the key SARS-CoV-2 spike protein that facilitates COVID-19 infection, studies have previously identified several such highly flexible regions with therapeutic importance. However, protein structures available from the Protein Data Bank are presented as static snapshots that may not adequately depict this flexibility, and furthermore these cannot keep pace with new mutations and variants. In this paper we present a sequential Monte Carlo method for broadly sampling the 3-D conformational space of protein structure, according to the Boltzmann distribution of a given energy function. Our approach is distinct from previous sampling methods that focus on finding the lowest-energy conformation for predicting a single stable structure. We exemplify our method on the SARS-CoV-2 omicron variant as an application of timely interest. Our results identify sequence positions 495-508 as a key region where omicron mutations have the most impact on the space of possible conformations, which coincides with the findings of other preliminary studies on the binding properties of the omicron variant.

stat.AP↗

arXiv2021

Comparing regional and provincial-wide COVID-19 models with physical distancing in British Columbia

We study the effects of physical distancing measures for the spread of COVID-19 in regional areas within British Columbia, using the reported cases of the five provincial Health Authorities. Building on the Bayesian epidemiological model of Anderson et al. (2020), we propose a hierarchical regional Bayesian model with time-varying regional parameters between March to December of 2020. In the absence of COVID-19 variants and vaccinations during this period, we examine the regionalized basic reproduction number, modelled prevalence, relative reduction in contact due to physical distancing, and proportion of anticipated cases that have been tested and reported. We observe significant differences between the regional and provincial-wide models and demonstrate the hierarchical regional model can better estimate regional prevalence, especially in rural regions. These results indicate that it can be useful to apply similar regional models to other parts of Canada or other countries.

stat.AP↗

arXiv2021

Multimodel Bayesian Analysis of Load Duration Effects in Lumber Reliability

This paper evaluates the reliability of lumber, accounting for the duration-of-load (DOL) effect under different load profiles based on a multimodel Bayesian approach. Three individual DOL models previously used for reliability assessment are considered: the US model, the Canadian model, and the Gamma process model. Procedures for stochastic generation of residential, snow, and wind loads are also described. We propose Bayesian model-averaging (BMA) as a method for combining the reliability estimates of individual models under a given load profile that coherently accounts for statistical uncertainty in the choice of model and parameter values. The method is applied to the analysis of a Hemlock experimental dataset, where the BMA results are illustrated via estimated reliability indices together with 95% interval bands.

stat.AP↗