Recent advancements in nuclear engineering have seen significant integration of machine learning (ML) technologies, resulting in enhanced design, safety, and operational efficiency of nuclear reactors. The emerging discipline of Scientific Machine Learning (SciML) combines traditional scientific computing with advanced ML techniques to address complex challenges in nuclear engineering. Notable applications include physics-informed machine learning, surrogate modeling, Bayesian inverse problems, and digital twins, which have improved the accuracy and efficiency of reactor simulations and predictive maintenance. One prominent area of development is the optimization of Small Modular Reactors (SMRs) and advanced fission reactors, such as molten salt and sodium-cooled reactors. These reactors benefit from ML algorithms that enhance safety features, operational efficiency, and cost-effectiveness by allowing real-time data analysis and anomaly detection. Additionally, ML is being used to streamline the licensing process for micro-reactors, which are compact and mobile, making them ideal for remote locations. In fusion energy, machine learning aids in controlling plasma behavior and optimizing fusion reactions, bringing us closer to achieving practical and sustainable fusion power. These technological advancements are supported by regulatory frameworks that adapt to the rapid evolution in nuclear technology, ensuring both safety and environmental compliance. Overall, the integration of machine learning in nuclear engineering represents a transformative step towards more resilient, efficient, and safe nuclear power systems, paving the way for a sustainable and carbon-free energy future.

nuclear reactors, Scientific Machine Learning (SciML), physics-informed machine learning, surrogate modeling, Bayesian inverse problems, digital twins