Uncertainty Quantification for Data-Driven Machine Learning Models in Nuclear Engineering Applications: Where We Are and What Do We Need?

2025-03-16

AI Summary

Uncertainties in data-driven models—arising from measurement noise, insufficient data coverage, extrapolation risks, and training stochasticity—severely hinder their trustworthy deployment in safety-critical nuclear engineering applications. Method: This work systematically contrasts the fundamental uncertainty quantification (UQ) paradigms of physics-based versus data-driven modeling for nuclear systems. It introduces a nuclear-engineering-specific UQ taxonomy and a unified verification–validation–UQ framework, integrating Bayesian deep learning, ensemble methods, sensitivity analysis, and coverage assessment. Empirical validation is conducted on canonical tasks: criticality safety prediction and fault diagnosis. Contribution/Results: The study identifies dominant uncertainty sources, establishes a reproducible, evaluable UQ implementation pathway with standardized benchmarks, and delivers both theoretical foundations and practical guidelines for credibility assessment of high-reliability nuclear AI systems.

Technology Category

Machine Learning: Calibration & Uncertainty QuantificationReasoning under Uncertainty: Uncertainty RepresentationsGame Theory and Economic Paradigms: Adversarial Learning

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingUser Modeling, Personalization and Recommendation: Fairness-aware retrieval and rankingGraph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Machine learning (ML) has been leveraged to tackle a diverse range of tasks in almost all branches of nuclear engineering. Many of the successes in ML applications can be attributed to the recent performance breakthroughs in deep learning, the growing availability of computational power, data, and easy-to-use ML libraries. However, these empirical successes have often outpaced our formal understanding of the ML algorithms. An important but under-rated area is uncertainty quantification (UQ) of ML. ML-based models are subject to approximation uncertainty when they are used to make predictions, due to sources including but not limited to, data noise, data coverage, extrapolation, imperfect model architecture and the stochastic training process. The goal of this paper is to clearly explain and illustrate the importance of UQ of ML. We will elucidate the differences in the basic concepts of UQ of physics-based models and data-driven ML models. Various sources of uncertainties in physical modeling and data-driven modeling will be discussed, demonstrated, and compared. We will also present and demonstrate a few techniques to quantify the ML prediction uncertainties. Finally, we will discuss the need for building a verification, validation and UQ framework to establish ML credibility.

Problem

• Quantify uncertainty in nuclear engineering ML models
• Compare UQ in physics-based and data-driven models
• Develop verification and validation for ML credibility

Innovation

• Uncertainty quantification for ML models
• Comparison of UQ in physics and data models
• Techniques to quantify ML prediction uncertainties