Abstract:In the era of big data, machine learning (ML) has become a powerful tool in various fields, notably impacting structural dynamics. ML algorithms offer advantages by modeling physical phenomena based on data, even in the absence of underlying mechanisms. However, uncertainties such as measurement noise and modeling errors can compromise the reliability of ML predictions, highlighting the need for effective uncertainty awareness to enhance prediction robustness. This paper presents a comprehensive review on navigating uncertainties in ML, categorizing uncertainty-aware approaches into probabilistic methods (including Bayesian and frequentist perspectives) and non-probabilistic methods (such as interval learning and fuzzy learning). Bayesian neural networks, known for their uncertainty quantification and nonlinear mapping capabilities, are emphasized for their superior performance and potential. The review covers various techniques and methodologies for addressing uncertainties in ML, discussing fundamentals and implementation procedures of each method. While providing a concise overview of fundamental concepts, the paper refrains from in-depth critical explanations. Strengths and limitations of each approach are examined, along with their applications in structural dynamic forward problems like response prediction, sensitivity assessment, and reliability analysis, and inverse problems like system identification, model updating, and damage identification. Additionally, the review identifies research gaps and suggests future directions for investigations, aiming to provide comprehensive insights to the research community. By offering an extensive overview of both probabilistic and non-probabilistic approaches, this review aims to assist researchers and practitioners in making informed decisions when utilizing ML techniques to address uncertainties in structural dynamic problems.

What problem does this paper attempt to address?

The core problem that this paper attempts to solve is: in structural dynamics, how can machine learning (ML) methods effectively handle and quantify uncertainty. Specifically, the paper aims to explore and summarize the applications of probabilistic and non - probabilistic methods in forward problems (such as response prediction, sensitivity assessment, and reliability analysis) and inverse problems (such as system identification, model updating, and damage identification) to improve the robustness and reliability of ML predictions. ### 1. Background and Motivation With the advent of the big data era, machine learning has become a powerful tool in the fields of science and engineering, especially in structural dynamics. However, due to the existence of uncertainties such as measurement noise and modeling errors, the reliability of ML predictions may be affected. Therefore, how to effectively handle these uncertainties and enhance the robustness of predictions has become an urgent problem to be solved. ### 2. Research Objectives The main objectives of the paper are: - **Review Uncertainty Handling Methods**: Conduct a comprehensive review of existing probabilistic and non - probabilistic uncertainty handling methods. - **Classification and Comparison**: Classify uncertainty handling methods into probabilistic (including Bayesian and frequentist) and non - probabilistic (such as interval learning and fuzzy learning) types, and discuss in detail the basic principles and implementation steps of each method. - **Application Areas**: Explore the specific applications of these methods in forward and inverse problems in structural dynamics. - **Research Gaps and Future Directions**: Identify the gaps in current research and propose future research directions. ### 3. Method Classification The paper mainly divides uncertainty handling methods into two major categories: - **Probabilistic Methods**: Represent the uncertainty of model predictions through probability distributions. This type of method is further subdivided into Bayesian methods and non - Bayesian methods. Bayesian neural networks have received special attention due to their superior performance in uncertainty quantification and nonlinear mapping. - **Non - Probabilistic Methods**: Adopt a different perspective from probabilistic methods to handle prediction uncertainty. For example, interval learning quantifies uncertainty through upper and lower bounds, while fuzzy learning reflects the uncertainty in classification or clustering results through membership functions. ### 4. Application Examples The paper also discusses the specific applications of these methods in structural dynamics, including but not limited to: - **Forward Problems**: Such as response prediction, sensitivity analysis, and reliability analysis. - **Inverse Problems**: Such as system identification, model updating, and damage identification. ### 5. Conclusions and Prospects Through a comprehensive review of existing literature, the paper not only summarizes the advantages and limitations of various uncertainty handling methods, but also points out the deficiencies in current research and provides directions for future research. This helps researchers and practitioners make more informed decisions when using ML techniques to solve structural dynamics problems. In conclusion, this paper is committed to providing a comprehensive uncertainty handling guide for machine learning applications in structural dynamics to improve the reliability and interpretability of model predictions.

Navigating Uncertainties in Machine Learning for Structural Dynamics: A Comprehensive Review of Probabilistic and Non-Probabilistic Approaches in Forward and Inverse Problems

A Structured Review of Literature on Uncertainty in Machine Learning & Deep Learning

Machine learning with data assimilation and uncertainty quantification for dynamical systems: a review

Machine Learning for Risk and Resilience Assessment in Structural Engineering: Progress and Future Trends

Data-Driven Uncertainty Quantification and Propagation in Structural Dynamics through a Hierarchical Bayesian Framework

Machine learning aided uncertainty quantification for engineering structures involving material-geometric randomness and data imperfection

Uncertainty Analysis for Non-identifiable Dynamical Systems: Profile Likelihoods, Bootstrapping and More

Recent Advances and Applications of Machine Learning in Experimental Solid Mechanics: A Review

Machine learning-based methods in structural reliability analysis: A review

Physics-Informed Machine Learning for Seismic Response Prediction OF Nonlinear Steel Moment Resisting Frame Structures

Towards a population-informed approach to the definition of data-driven models for structural dynamics

Applying Machine Learning Techniques: Uncertainty Quantification in Nonlinear Dynamics Characters Predictions via Gated Recurrent Unit-Based Reduced-Order Models

Machine learning for structural engineering: A state-of-the-art review

Towards Using Probabilistic Models to Design Software Systems with Inherent Uncertainty

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

Machine Learning Techniques in Structural Wind Engineering: A State-of-the-Art Review

Learning Latent Space Dynamics with Model-Form Uncertainties: A Stochastic Reduced-Order Modeling Approach

Physics-informed machine learning and its structural integrity applications: state of the art

A Machine Learning Approach to Model Interdependencies between Dynamic Response and Crack Propagation