Over the past few decades, computational models have become indispensable tools in engineering and applied sciences for the analysis and design of complex systems. High-fidelity simulators, such as finite element or multiphysics models, are now routinely employed for tasks including uncertainty propagation, reliability and robustness assessment, optimization under uncertainty, Bayesian inference, and global sensitivity analysis. However, the direct use of such simulators often remains computationally prohibitive, even with modern high-performance computing resources. This challenge becomes even more pronounced when the underlying simulators are stochastic, i.e., when repeated runs at fixed input parameters produce different outputs due to internal randomness or hidden latent variables.

To address these limitations, surrogate models have emerged as powerful fast-to-evaluate approximations of complex simulators. This mini-symposium aims to highlight recent advances and emerging trends in surrogate modelling for both deterministic and stochastic computational models. Contributions are invited on established approaches such as (sparse) polynomial chaos expansions, Gaussian process regression/Kriging, support vector regression, sparse-grid interpolation, and stochastic reduced-order models, as well as modern machine learning techniques including deep neural networks, ensemble learning, random forests, diffusion models, flow matching / normalizing flows. Topics of interest include active learning, dimensionality reduction, multi-fidelity modelling, quantification of surrogate-induced errors, and surrogate modelling in high-dimensional or stochastic settings. Applications in engineering and applied sciences that demonstrate the capabilities and limitations of surrogate methods are especially welcome.

Surrogate models (aka metamodels, emulators), i.e., computationally inexpensive approximations of high-fidelity numerical models, are routinely used in computational science and engineering applications to accelerate or simply enable costly numerical studies like optimization and uncertainty quantification. The class of data-driven surrogate models computed by means of machine learning (ML) methods have become quite popular for such tasks. Next to approximation accuracy and computational speed, the concept of predictive uncertainty quantification for surrogate models is getting traction. It is increasingly recognized that information about the uncertainty in the surrogate model’s predictions is often as important as the prediction itself, especially considering surrogate-based estimates that may affect safety, cost, or scientific conclusions.

This minisymposium aims to cover methodological developments in uncertainty-aware surrogate models, i.e., surrogate models that accompany their predictions with confidence or predictive intervals, as well as relevant applications in computational science and engineering. Topics of interest include (but are not limited to):

• Bayesian models, e.g., Gaussian processes and Bayesian neural networks.
• Frequentist approaches, e.g., conformal prediction and bootstrapping/bagging.
• Challenges due to high-dimensional model inputs and outputs.
• Uses in active learning, reliability analysis, and optimization under uncertainty.
• Science and engineering applications showcasing practical insights.

Keywords:

predictive uncertainty quantification; data-driven surrogate models; uncertainty-aware surrogate models; conformal prediction; Bayesian neural networks; Gaussian process regression; model ensembles

The ability to make decisions under limited data is becoming increasingly important in the context of modelling for engineering applications. Several approaches are currently emerging to perform Uncertainty Quantification (UQ) in models of complex systems, structures, and components. These methods range from purely interval and fuzzy approaches to advanced probabilistic schemes and imprecise probabilistic concepts. In addition, the application of machine learning-inspired techniques, such as active learning, is becoming ever more popular in this context. This mini-symposium focuses on novel developments of these techniques for the representation of uncertainty and applications in advanced engineering modelling activities, especially in the context of UQ, reliability analysis and sensitivity analysis.

Researchers and practitioners focusing on UQ, reliability analysis and sensitivity analysis in numerical modelling for engineering applications with limited data are invited to submit an abstract to this mini-symposium. Topics include, but are not limited to, uncertainty propagation methodologies, inverse identification and quantification techniques, optimisation under uncertainty, as well as recent implementations and developments of Machine Learning for dealing with uncertainty.

Engineering systems across multiple disciplines are characterized by complex, high-dimensional, and deeply uncertain environments. Traditional reliability analysis methods, while powerful, often rely on assumptions regarding probabilistic completeness, model fidelity, and separability of uncertainty sources that may not hold in modern applications. This mini-symposium aims to bring together researchers working on robust reliability analysis under deep uncertainty, with a particular focus on methodological advances and cross-disciplinary perspectives. The scope of the session includes (but is not limited to):

• Reliability analysis under epistemic uncertainty;
• Robust and risk-informed decision-making;
• Rare-event simulation techniques;
• Surrogate and reduced-order modeling for uncertainty quantification.
• Special emphasis is placed on methods that enhance the robustness, interpretability, and computational efficiency of reliability assessments in complex engineering systems.

This mini-symposium is organized under the auspices of the Technical Committee on Robust Reliability, Risk & Decision Analysis (TC R3) of International Association for Structural Safety & Reliability (IASSAR). It aims to strengthen the link between the uncertainty quantification community and structural reliability research, fostering dialogue between methodological developments and engineering applications. We particularly encourage contributions that bridge theoretical advances with computational methods and real-world applications in structural engineering, structural dynamics, and multi-physics systems. The session will provide a platform for discussion on emerging challenges and future directions in robust reliability analysis.

Recent advances in generative AI, scientific machine learning, and uncertainty quantification are redefining how complex scientific and engineering systems are represented, simulated, and controlled. Classical workflows based on explicit governing equations and high-fidelity simulation remain indispensable, but they often become computationally prohibitive in regimes characterized by high dimensionality, multiphysics coupling, multiscale dynamics, nonlinear behavior, and entangled aleatory and epistemic uncertainties. This minisymposium focuses on next-generation generative modeling paradigms for scientific computing, including diffusion and score-based models, variational and flow-based architectures, latent-variable models, neural operators, and geometry-aware generative methods. These approaches offer powerful mechanisms for learning structured latent representations, sampling from high-dimensional probability measures, discovering low-dimensional manifolds, accelerating surrogate and reduced-order modeling, and quantifying uncertainty in data-scarce or simulation-limited regimes. Particular emphasis will be placed on physically consistent and uncertainty-aware generative models that integrate domain knowledge, geometric structure, conservation principles, and probabilistic calibration. Topics of interest include generative modeling on manifolds, data-efficient learning, probabilistic surrogate modeling, and generative approaches for scientific discovery and engineering design. The MS aims to bring together researchers developing theory, algorithms, and applications at the interface of generative AI, UQ, and scientific machine learning, with the goal of advancing reliable, interpretable, and computationally efficient AI-enabled modeling of complex systems.

Keywords: generative AI; score-based diffusion models; manifold learning, probabilistic learning; uncertainty quantification

Engineering systems and structures are increasingly required to operate under complex uncertainties arising from stochastic excitations, material and geometric variability, sparse data, model-form errors, degradation processes, and changing environmental conditions. Reliable prediction and risk-informed decision-making therefore require computational methods that can combine mechanistic knowledge, probabilistic modelling, data-driven learning, and uncertainty quantification in a coherent and efficient manner.

This mini symposium aims to bring together recent advances in physics-enhanced and data-driven methods for uncertainty quantification, stochastic dynamics, structural reliability, and risk-informed engineering analysis. Particular attention will be given to methods that integrate physical laws, computational mechanics, stochastic modelling, and machine learning to improve predictive reliability under limited, noisy, or heterogeneous data. Relevant topics include, but are not limited to:

• physics-informed and physics-enhanced machine learning for stochastic dynamic systems;
• surrogate and reduced-order modelling for uncertainty propagation;
• Bayesian and active learning methods for reliability analysis;
• probability density evolution and stochastic simulation;
• hybrid aleatory-epistemic uncertainty quantification;
• multi-fidelity modelling, rare-event probability estimation, time-dependent reliability, and risk-informed assessment of engineering systems.

The mini symposium welcomes both methodological contributions and engineering applications, particularly in structural dynamics, earthquake engineering, civil infrastructure, mechanical and aerospace systems, energy systems, digital twins, resilient infrastructure, and computational mechanics under uncertainty. The goal is to provide a focused forum for discussing how physics-based modelling and data-driven learning can be combined to achieve credible, interpretable, and computationally efficient uncertainty quantification for complex engineering systems.

Bayesian analysis in science and engineering addresses the fundamental challenge of reliably modelling and predicting the behavior of systems under uncertainty by combining models with data. In real-world applications, this task is complicated by complex system dynamics, limited and noisy data, and the high dimensionality of uncertain model parameters. Addressing these challenges requires advanced methods for quantifying uncertainties and calibrating models based on observed data. This minisymposium aims to bring together researchers working on the theoretical development and practical application of Bayesian methods in uncertainty quantification, propagation and decision-making. We invite contributions exploring theoretical or computational aspects, such as sampling strategies and approximation techniques for solving Bayesian inverse problems, and their application to uncertainty quantification in physics-based and data-driven models. We also welcome work that advances the propagation of quantified uncertainties through complex systems and supports robust, data-informed decision-making under uncertainty.

Complex heterogeneous materials offer unique combinations of mechanical, thermal, and functional properties resulting from their intricate microstructures. Accurate multiscale modelling is key to understanding their behavior, predicting their performance, and enabling their optimization in advanced applications [1]. In recent years, major advances in deterministic and stochastic methods have made multiscale modelling an expanding research area [2]. This Mini symposium will highlight recent progress in advanced analytical and computational tools for studying complex heterogeneous media, with emphasis on rigorous deterministic approaches and developments in stochastic and data-driven modelling. Topics of interest include, but are not limited to:

• Advanced analytical and computational methods for deterministic and stochastic modelling of periodic/random microstructures and fracture/damage evolution
• Random field modelling of heterogeneous media
• Design and optimization of composite materials and structures under uncertainty
• Scale-bridging techniques, including analytical and computational homogenization and multiscale finite element methods
• Multiphysics modelling of heterogeneous materials, including coupled mechanical, chemical, thermal, and electrical processes
• Machine learning, data-driven techniques, and high-performance computing for efficient multiscale material analysis

Keywords: Complex materials, Multiscale and Multiphysics modelling, Analytical and computational tools.

Acknowledgement
The organizers acknowledge the grant FISA (CUP: B83C25000810001) and  Sapienza Ateneo Progetti Grandi (CUP: B83C24007070005).

REFERENCES

[1] T. Sadowski and P. Trovalusci, Multiscale Modeling of Complex Materials, 556, Springer, 2014.
[2] G. Stefanou, D. Savvas and M. Papadrakakis, Stochastic finite element analysis of composite structures based on mesoscale random fields of material properties, Comput. Methods Appl. Mech. Eng. 326 (2017) 319-337.

Transport phenomena arise in a wide range of applications in engineering and the sciences, and can be modeled at varying levels of resolution. Common approaches include macroscopic fluid-dynamical models, mesoscopic kinetic descriptions, and microscopic particle methods. While numerical methods are available for the classical simulation of such problems across different resolutions, many open challenges remain in combining these methods with uncertainty quantification. In multi-resolution settings, suitable models for uncertainty quantification have to be selected balancing computational cost, modeling error, and simulation error to achieve the best possible results. In this minisymposium, we bring together researchers working on uncertainty quantification for such transport phenomena to present state-of-the-art methods, discuss recent advances, and exchange ideas.

 Physics-informed machine learning has emerged as a powerful paradigm for combining engineering judgement with operational measurements. Purely data-driven models often struggle in engineering settings, where labelled data are scarce, operating conditions vary, and predictions must remain physically plausible to be trusted. Conversely, purely physics-based models can be costly to evaluate and may fail to capture the complexity of real-world behaviour. By embedding physical knowledge, governing equations, conservation laws, or known constraints, into machine learning models, physics-informed approaches offer improved generalisation, data efficiency, and interpretability, making them well suited to safety-critical engineering applications.

This minisymposium considers how physics-informed machine learning can support improved decision-making across the lifecycle of complex engineering systems, from design and monitoring through to maintenance and asset management. We welcome contributions spanning new methodological developments, principled treatment of uncertainty, and demonstrations on real engineering structures and infrastructure. A particular emphasis is placed on how the integration of data and physics can translate into actionable, trustworthy decisions.

Topics include, but are not limited to:

• New methods in physics
• informed machine learning for engineering systems
• Uncertainty quantification in physics
• informed machine learning
• Digital twins for asset management and decision-making
• Population-level physics-informed machine learning
• Physics-informed methods for structural health monitoring and prognostics

Computational models of engineering and scientific systems inevitably contain parameters whose values are uncertain, expensive to measure, or only accessible through indirect observation. Global sensitivity analysis (GSA) has established itself as a principled framework for ranking the influence of these parameters on model outputs, supporting model reduction, robustness assessment, and validation. Yet the scope of GSA extends well beyond parameter ranking: at its core, every sensitivity index is a statement about how much information an input carries about an output, and this information-theoretic perspective opens a richer set of questions about how models and data interact.

This mini-symposium invites contributions that explore sensitivity analysis through the lens of information gain and information loss. Rather than treating information gain as a by-product of sensitivity analysis, we propose it as an organizing principle: the goal of the analysis is not only to identify which parameters matter, but to quantify what can be learned about parameters, model structure, and the system being modeled. The complementary concept of information loss is equally central: how much information is lost when a high-fidelity model is replaced by a surrogate, when a parameter is fixed, or when an experimental design is suboptimal? The chain connecting derivative-based measures to total-order Sobol indices and to the Fisher Information Matrix provides a concrete theoretical backbone for both questions. This framing connects GSA to a cluster of related problems that have so far been addressed in relative isolation: optimal experimental design, Bayesian model calibration, value of information analysis, identifiability analysis, and the interpretation of data-driven and hybrid models. By foregrounding information gain and loss, the mini-symposium aims to bring these threads into dialogue and to develop a more unified view of how sensitivity analysis supports knowledge acquisition across the model development lifecycle.

Beyond the individual presentations, the mini-symposium aims to foster a broader discussion around a unified information-theoretic perspective on sensitivity analysis. Depending on the interest of participants and the outcomes of the session, this discussion may provide the basis for a future community position or white paper on information gain, information loss, and knowledge acquisition across the model development lifecycle.

The computational analysis of heterogeneous materials with uncertain microstructures often requires extensive microscale simulations to quantify the influence of geometrical, topological, and material uncertainties on the effective macroscopic response. Such simulations become prohibitively expensive when uncertainty quantification, reliability assessment, optimization, inverse identification, or design studies are involved.

Recent advances in surrogate modelling, reduced-order modelling, machine learning, and data-driven computational mechanics offer new opportunities to accelerate computational multiscale analyses while maintaining predictive accuracy. Surrogate models can efficiently approximate the complex relationships between uncertain microstructural features and effective material behavior, enabling large-scale uncertainty quantification, robust design, and real-time predictions.

This Minisymposium aims to foster interaction and exchange in the emerging field of surrogate-assisted computational multiscale modelling of heterogeneous materials with uncertain microstructures. Contributions addressing probabilistic and non-probabilistic uncertainties in material microstructures are particularly welcome. Relevant topics include, but are not limited to:

• Machine learning for computational homogenization
• Surrogate models for multiscale materials modelling
• Reduced-order modelling of heterogeneous materials
• Data-driven constitutive modelling under uncertainty
• Active learning and adaptive sampling strategies
• Surrogate modelling for multiphysics and multiscale systems
• AI-enhanced computational homogenization

Contributions to all these topics are kindly welcomed in this Minisymposium.

Due to factors related to manufacturing or construction processes, ageing, loading, environmental & boundary conditions, measurement errors, modeling assumptions / inefficiencies and numerous others, almost every engineering system is characterized by uncertainty. The propagation of uncertainty through the system gives rise to corresponding complexities during simulation of its structural response, yet also during its characterization based on experimental data. Consequently, only a limited degree of confidence can be attributed in the behavior, reliability and safety of structural systems in particular throughout their life cycle. For this purpose, it is imperative to develop models and processes able to encompass the aforementioned uncertainties. 

This mini-symposium deals with uncertainty quantification and propagation methods applicable to the simulation and identification of complex engineering systems. It covers theoretical and computational issues, applications in structural dynamics, earthquake engineering, mechanical and aerospace engineering, as well as other related engineering disciplines. Topics relevant to the session include: dynamics of structural systems, structural health monitoring methods for damage and reliability prognosis, theoretical and experimental system identification for systems with uncertainty, uncertainty quantification in model selection and parameter estimation, stochastic simulation techniques for state estimation and model class selection, structural prognosis techniques, updating response and reliability predictions using data. Papers dealing with experimental investigation and verification of theories are especially welcomed.

Modern engineering systems are increasingly required to operate safely and efficiently under complex and uncertain environments throughout their service life. Aging infrastructure, extreme climatic events, variable operational loads, material degradation, manufacturing imperfections, and limited sensing information introduce significant uncertainties that challenge traditional approaches to structural assessment and maintenance. While physics based models provide valuable insights into the behavior of engineered systems, they often suffer from modeling simplifications and inherent uncertainty. Conversely, data-driven methods can exploit the sensor and monitoring data but may lack physical interpretability, robustness, and extrapolation capabilities. Addressing these limitations requires a new generation of methodologies that effectively integrate physical knowledge with observational data to enable reliable uncertainty quantification, damage detection, condition assessment, and predictive maintenance.

This mini-symposium aims to bring together researchers and practitioners working at the intersection of computational science, structural health monitoring (SHM), uncertainty quantification (UQ), machine learning, and digital twin technologies. The focus is on innovative frameworks that combine governing physical principles, numerical simulations, experimental observations, and real-time sensor data to improve the monitoring, diagnosis, prognosis, and management of engineering systems operating under uncertainty.

Topics of interest include, but are not limited to:

Physics-informed machine learning and scientific computing for SHM and UQ

Digital twins and hybrid physics-data modeling frameworks • Surrogate modeling and reduced-order modeling for UQ • Operator learning for UQ and SHM • Bayesian inference, data assimilation, and model updating • Probabilistic structural dynamics and reliability analysis • Sensor fusion and multi-modal data integration • Damage detection, localization, and prognosis • Uncertainty-aware decision-making and risk-informed maintenance • Real-time monitoring and forecasting of engineering systems • Transfer learning, domain adaptation, and federated learning for SHM applications • Applications involving bridges, buildings, wind turbines, aircraft, spacecraft, offshore platforms, mechanical systems, and advanced manufacturing systems The mini-symposium seeks to foster interdisciplinary discussions on emerging technology, computational methodologies, and real-world applications that leverage the fusion of data and physics to enhance the reliability, resilience, safety, and sustainability of next-generation engineered systems.

Surrogate models have established themselves as an increasingly common and powerful tool for classical engineering scenarios, enabling efficient uncertainty quantification and optimization in steady-state applications. However, extending this success to time-dependent systems remains a significant challenge. Unlike deterministic steady-state models, time-dependent simulators (such as those describing structural dynamics, environmental time-series, or complex multi-physics evolution) exhibit inherently high-dimensional and computationally demanding behavior, often rendering traditional uncertainty quantification (UQ) methods intractable. There is a pressing need for accurate, efficient surrogate models capable of capturing these dynamical behaviors.

This mini-symposium aims to facilitate discussion on advanced surrogate modeling techniques specifically tailored for time-dependent problems. We invite contributions covering the state-of-the-art in this field, including but not limited to autoregressive models, Long Short-Term Memory (LSTM) networks, neural operators, and other advanced machine learning techniques capable of accelerating UQ- and System Identification- related calculations for both stationary and non-stationary time-dependent simulators and datasets.  

The integration of Uncertainty Quantification (UQ) is increasingly recognized as a fundamental component of robust, quantitative, and performance-based engineering. This trend is increasingly supported by the development of comprehensive software ecosystems that facilitate access to high-fidelity analysis, reducing the computational requirements and technical specialized knowledge previously needed to manage complex simulations in the presence of uncertainty.

Modern UQ software supports a diverse community, including researchers developing new algorithms and industry practitioners implementing UQ in production workflows. Ensuring the portability, reproducibility, and integration of novel tools and algorithms is essential for the dissemination of uncertainty quantification and management techniques across different disciplines.

This mini-symposium aims to facilitate discussion between established researchers and new developers in the international UQ software community. We intend to bridge the gap between software developers and users in various research and development sectors. We invite contributions on topics such as advanced non-intrusive UQ methods, surrogate modeling using machine learning, scalable HPC workflows, general-purpose UQ frameworks, and case studies of UQ applications to industrial challenges. 

Uncertainty quantification (UQ) enhances the reliability and credibility of model-based predic tions, which is essential for informed decision-making and effective risk management across science and engineering. Classical UQ methods, including those based on Monte Carlo simulations, require mul tiple model evaluations to achieve accurate uncertainty estimates. Engineering models often rely on costly numerical solutions of partial differential equations, making the generation of numerous model evaluations highly computationally expensive and resource demanding. Model reduction addresses this problem by identifying dominant subspaces and constructing surrogate models to approximate the behavior of the full model through projection of the high-dimensional model onto a low-dimensional subspace. Unlike purely data-driven surrogate models, projection-based model reduction delivers surro gates that reflect the structure of the governing equations of the underlying physical system, facilitating analysis of the reduced models and often enabling accurate approximation with zero or limited data.

We invite talks that discuss methodological developments and novel applications of model reduc tion in all areas of uncertainty quantification, including but not limited to uncertainty propagation, uncertainty-based sensitivity analysis, Bayesian inversion, reliability analysis and optimization under uncertainty.

Scientific machine learning is rapidly transforming computational science and engineering by enabling data-driven surrogate models, accelerating large-scale simulations, and supporting real-time decision-making. However, many of these approaches still rely on deterministic learning paradigms that do not explicitly account for uncertainty, limiting their reliability in predictive and safety-critical applications.

This minisymposium focuses on uncertainty as a fundamental component of scientific machine learning, rather than as an afterthought. We aim to bring together researchers working at the intersection of machine learning, computational physics, and uncertainty quantification to develop methods that are both expressive and trustworthy.

Topics of interest include, but are not limited to, probabilistic neural operators and operator learning under uncertainty; Bayesian neural networks for inverse and forward problems; uncertainty-aware surrogate modeling for high-dimensional and multiscale systems; active learning and adaptive sampling strategies for efficient data acquisition; and the integration of uncertainty-aware models within digital twin frameworks for monitoring, prediction, and control. By emphasizing principled uncertainty representation in scientific machine learning, this minisymposium seeks to foster dialogue between communities in computational mechanics, applied mathematics, statistics, and machine learning, and to promote robust methodologies for data-driven scientific discovery.

Uncertainty quantification (UQ) and Machine Learning (ML) are two up to date and cutting -edge theory and technology in Sciences and Engineering. UQ provides a numerical relation of the uncertainty of the output response of a physical system when the respective variability of the input is defined. ML is the tool to interconnect the data driven approach to real problem solutions. Namely, the Ambrosios Antonios Savvides dataset obtained that govern a physical problem is processed and functions-models are formulated that provide the solution and response to the aforementioned problem. UQ and ML can be combined in order to obtain a broader solution to the physical problem considering its stochastic nature.

In this minisymposium, all applications for all possible aspects in Sciences and Engineering, of all Engineering disciplines, that implement UQ and ML are welcomed. Some representative but not restrictive occasions are provided.

1. UQ studies that implement Stochastic Random Field sums (Karhunen-Loeve, Spectral Representation), importance or fair probability sampling (Latin Hypercube sampling, Markov Chain Monte Carlo simulations)
2. ML studies that use all possible aspects of the theory (Neural Networks, optimization methods and Neural Networks formulation, convolutional autoencoders, methods of modelling with small dataset size)
3. All possible theoretical works that at least have a small application are also welcomed

Any inquiries about the suitability of a work to this minisymposium, can be sent to the email provided.  

Predicting the behaviour and ensuring the reliability of dynamical and time-variant systems are of fundamental importance across a wide range of engineering disciplines. In practice, system responses are inherently uncertain due to stochastic excitations and parameter variability. These uncertainties interact with the underlying system dynamics, posing significant challenges for prediction, reliability assessment, and engineering decision-making. As engineering systems become increasingly complex and data-rich, there is a growing need for efficient and reliable computational methods to accurately characterize uncertainty and quantify the probability of rare but critical events. This symposium aims to bring together researchers and practitioners working on theoretical, computational, and data-driven approaches for the reliability analysis of dynamical and time-variant systems. Contributions addressing fundamental developments, numerical methodologies, and practical engineering applications are all welcome. Topics of interest include, but are not limited to:

• Reliability analysis of stochastic dynamical systems;
• First-passage and rare-event reliability analysis;
• Bayesian inference, data assimilation, and online model updating;
• Surrogate modelling and uncertainty quantification;
• Model reduction for dynamical systems;
• Reliability sensitivity analysis;
• Reliability-based and robust design optimization;
• Stochastic optimal control and decision-making under uncertainty;
• Time-dependent probability density evolution and response characterization;
• Probabilistic digital twins for engineering systems;
• Physics-informed and data-driven methods for dynamical systems.

We particularly encourage contributions that develop innovative theoretical frameworks, scalable computational algorithms, and interdisciplinary applications in civil, aerospace, mechanical, automotive, energy, and related engineering fields.

Modern infrastructure is increasingly characterized as systems-of-systems with strong interdependencies across physical, cyber, social, and organizational domains. Transportation, energy, water, communication, industrial, and offshore infrastructures are no longer isolated systems but interconnected systems-of-systems, whose performance depends on complex hierarchical interactions across multiple spatial and temporal scales. These interdependencies create new challenges for uncertainty quantification, reliability and robustness assessment, resilience evaluation, and risk-informed decision-making, particularly under multiple hazards and cascading failures.

Recent advances in probabilistic modeling, Bayesian inference, digital twins, machine learning, network science, and computational mechanics have enabled more realistic representations of uncertainty propagation in complex interconnected systems. Nevertheless, major challenges remain in integrating multiscale hierarchical models, quantifying epistemic and aleatory uncertainties, modeling cascading effects, accounting for dynamic interactions among subsystems, and supporting resilience-oriented decision making over the life cycle of infrastructure systems.

This mini-symposium aims to bring together researchers working on uncertainty quantification, structural reliability, network science, systems engineering, and infrastructure resilience to discuss recent methodological developments and engineering applications. Contributions addressing both theoretical advances and real-world case studies are encouraged.

Incomplete list of potential topics

• Uncertainty quantification in interdependent infrastructure systems;
• Reliability and robustness of systems-of-systems and hierarchical networks;
• Cascading failures and multiscale uncertainty propagation;
• Bayesian updating and digital twins for interconnected systems;
• Physics-informed and probabilistic machine learning;
• Multi-hazard and climate-resilient infrastructure;
• Network reliability and transportation systems;
• Cyber-physical infrastructure systems;
• Infrastructure resilience and life-cycle performance;
• Risk-informed decision-making under deep uncertainty;
• Hierarchical probabilistic graphical models;
• Optimal inspection, maintenance, and recovery planning;
• Multi-objective optimization of resilient infrastructure systems;

Uncertainties due to variations in material properties, geometric parameters, external loads, and environmental conditions can significantly influence the structural response. Incorporating these uncertainties into structural design optimization enables more informed decision-making by accounting for both the robustness and the performance of the optimized structures. However, data and information are characterized by both aleatory uncertainty, representing inherent variability, and epistemic uncertainty, resulting from incomplete knowledge and imprecision. To achieve realistic structural safety and risk assessments, both types of uncertainty should be incorporated into structural design optimization using probabilistic models based on random variables as well as non-probabilistic methods, including interval variables, fuzzy variables, and polymorphic (hybrid) uncertainty models. The computational design optimization of complex structures accounting for uncertainties requires a computationally expensive nested loop algorithm and needs ongoing research on efficient numerical methods. The scope of the minisymposium includes, but is not limited to, the following topics in the context of structural design optimization:

• Uncertainty quantification using probabilistic (random variables) and non probabilistic and non-traditional methods (interval /fuzzy variables, and polymorphic / hybrid / mixed uncertainty models)
• Surrogate modeling strategies
• Random fields and processes
• Performance-, robustness-, and resilience-based design
• Sampling methods
• Sensitivity analysis
• Risk analysis

The minisymposium addresses recent developments and current challenges in optimization under uncertainty, particularly in the context of structural mechanics applications, e.g., in solid mechanics, material modeling, stability analysis, structural dynamics, multi-scale and multi-physics simulation. It is organized under the auspices of the Technical Committee on Robust Reliability, Risk and Decision Analysis of the International Association for Structural Safety and Reliability (IASSAR).