Computing Research Group

Research Focus Areas

Explainable and Interpretable AI

We develop and apply techniques that enhance model transparency, including post-hoc explainability and symbolic learning, with a focus on high-stakes domains such as nuclear energy.

Large Language Models

We utilize cutting-edge open-source LLMs for natural language processing, investigating inherent biases, linguistic nuances, prompt engineering, and the interpretation of elements like sarcasm and emojis.

Social Media Computing

We deploy language models to power real-time visualization dashboards that track and analyze public sentiment toward clean energy technologies—such as nuclear and renewable energy—across various social media platforms.

Uncertainty Quantification

We apply Bayesian calibration to refine model accuracy and use variational inference techniques for real-time data assimilation and uncertainty estimation.

Generative AI

We employ diffusion mode ls to generate synthetic datasets for engineering applications and use text-to-image models to create visually compelling and accurate representations of complex systems.

Recent Publications

Efficient sampling of polycyclic aromatic compounds for free energy predictions through active learning

2025

Energy and AI

View Paper

Fairness and social bias quantification in Large Language Models for sentiment analysis

2025

Knowledge-Based Systems

View Paper

Data efficiency assessment of generative adversarial networks in energy applications

2025

Energy and AI

pyMAISE: A Python platform for automatic machine learning and accelerated development for nuclear power applications

2024

Progress in Nuclear Energy

A comparative analysis of text-to-image generative AI models in scientific contexts: a case study on nuclear power

2024

Scientific Reports

Recent Funded and Ongoing Projects

•Development of real-time visualization dashboards for analyzing public opinion on clean energy technologies via social media
(Further read:
Work in progress
)
•Leveraging large language models to analyze public sentiment on nuclear power using social media and targeted surveys
(Further read: https://doi.org/10.1016/j.rser.2024.114570)
•Creating domain - specific text-to-image generative models for accurate scientific visualizations of nuclear power systems
(Further read: https://www.nature.com/articles/s41598-024-79705-4)
•Evaluating data requirements for machine learning and generative AI in engineering applications
(Further read: https://doi.org/10.1016/j.egyai.2025.100501)
•Development of pyMAISE: An automated machinel earning framework with nuclear power benchmarks
(Further read: https://doi.org/10.1016/j.pnucene.2024.105568)
•Enhancing data availability and model accuracy in sustainable aviation fuel research using Bayesian methods and diffusion models
(Further read: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5211901)
•Surrogate modeling for optimization, sensitivity analysis, and uncertainty quantification in complex simulations
(Further read: https://doi.org/10.1016/j.ress.2019.106731)
•Calibrating the liquid mercury spallation target model in particle accelerators using evolutionary neural networks and Bayesian uncertainty quantification
(Further read: https://doi.org/10.1016/j.rinp.2022.105414)
•Bayesian model averaging for handling model-form uncertainty and propagating uncertainty in nuclear simulation codes
(Further read: https://doi.org/10.1016/j.ress.2019.04.020)
•Design optimization under uncertainty for advanced energy and spallation neutron systems
(Further read: https://doi.org/10.1016/j.ijhydene.2019.11.046)