![[2602.19584] Interpolation-Driven Machine Learning Approaches for Plume Shine Dose Estimation: A Comparison of XGBoost, Random Forest, and TabNet](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19584] Interpolation-Driven Machine Learning Approaches for Plume Shine Dose Estimation: A Comparison of XGBoost, Random Forest, and TabNet
Summary
This article compares interpolation-driven machine learning approaches for plume shine dose estimation, evaluating XGBoost, Random Forest, and TabNet models for accuracy and interpretability.
Why It Matters
The study addresses the critical need for accurate radiation dose assessment in nuclear safety, leveraging machine learning to improve prediction efficiency. By comparing different ML models, it highlights the potential of advanced techniques to enhance safety protocols in radiological emergency responses.
Key Takeaways
- Interpolation-assisted ML frameworks improve dose estimation accuracy.
- XGBoost outperforms Random Forest and TabNet in predictive accuracy.
- Tree-based models prioritize geometry-dispersion features over radionuclide identity.
- A web-based GUI facilitates interactive scenario evaluations.
- The study underscores the importance of high-resolution datasets in machine learning applications.
Computer Science > Machine Learning arXiv:2602.19584 (cs) [Submitted on 23 Feb 2026] Title:Interpolation-Driven Machine Learning Approaches for Plume Shine Dose Estimation: A Comparison of XGBoost, Random Forest, and TabNet Authors:Biswajit Sadhu, Kalpak Gupte, Trijit Sadhu, S. Anand View a PDF of the paper titled Interpolation-Driven Machine Learning Approaches for Plume Shine Dose Estimation: A Comparison of XGBoost, Random Forest, and TabNet, by Biswajit Sadhu and 3 other authors View PDF Abstract:Despite the success of machine learning (ML) in surrogate modeling, its use in radiation dose assessment is limited by safety-critical constraints, scarce training-ready data, and challenges in selecting suitable architectures for physics-dominated systems. Within this context, rapid and accurate plume shine dose estimation serves as a practical test case, as it is critical for nuclear facility safety assessment and radiological emergency response, while conventional photon-transport-based calculations remain computationally expensive. In this work, an interpolation-assisted ML framework was developed using discrete dose datasets generated with the pyDOSEIA suite for 17 gamma-emitting radionuclides across varying downwind distances, release heights, and atmospheric stability categories. The datasets were augmented using shape-preserving interpolation to construct dense, high-resolution training data. Two tree-based ML models (Random Forest and XGBoost) and one deep learning (D...