ML24263A255
| ML24263A255 | |
| Person / Time | |
|---|---|
| Issue date: | 09/17/2024 |
| From: | Doug Eskins Office of Nuclear Regulatory Research |
| To: | |
| References | |
| Download: ML24263A255 (1) | |
Text
Characterizing Nuclear Cybersecurity Using AI/ML
NRC AI Workshop Sept. 17, 2024 Doug Eskins, NRC Project Start: Oct 2022 Project Finish: May 2024 Project Team
NRC
- Doug Eskins (nuclear, M&S*)
- Kaitlyn Cottrell (AI/ML)
- Anya Kim (cybersecurity)
https://engineering.purdue.edu/NE/news/2023/us-nrc-Purdue University visits-the-school-of-nuclear-engineering
- Prof. Stylianos Chatzidakis (Dir Nuclear Engineering Radiation Lab)
- 5 graduate students (nuclear, AI, M&S)
- Modeling and Simulation Project Overview
Identify Data Collection Performance Develop &
Nuclear AI- & Model Evaluation & Document CyS Use CaseTrainingGap Analysis Insights
OBJECTIVES Cyber Events (1) Explore the use of AI/ML to classify Off Normal StatesOther nuclear cybersecurity Events states (2) Develop insights for Normal StatesNormal Operations assessment of nuclear AI-CyS Challenges and Insights Example event progression
- Real data - Data artifacts
- Algorithm performance - Robustness
- Data Management - Explainability
40000 Overall Null Values 35000 30000 25000 20000 15000 10000 5000 0
0 20 60 8040 Features Composite Classifier Reports
- Research Plan Development (TLR-RES/DE-2024-003a), ML23040A168
- Identification of a Representative Use Case (TLR-RES/DE-2024-003b),
- Identification of AI-ML Technology (TLR-RES/DE-2024- 003c),
- Use Case Implementation (TLR-RES/DE-2024- 003d), ML24052A002
- Performance Evaluation and Gap Analysis (TLR-RES/DE-2024- 003e),
- Final Report: Characterizing Nuclear Cybersecurity States using AI-ML (TLR-RES/DE-2024- 003), ML24193A008