Foundation models, signal extraction, and the learning systems that sit inside a tokamak's control loop.
Large neural networks trained on heterogeneous tokamak data, specialized to downstream diagnostic and control tasks. APS DPP 2025; ICLR 2026 Workshop DATA-FM.
Fast signal extraction for fluctuating tokamak time series. Real-time, fluctuation-aware feature pipelines for plasma diagnostics. arXiv Feb 2026.
Class-conditional normalizing flows as oracles for exact posteriors. Scaling laws, active learning, and the limits of learning from exact posteriors. arXiv Jan 2026.