Applied Electromagnetics LabResearch question
How can field physics, numerical simulation, uncertainty, and data-driven models be combined without losing physical interpretability?
From electromagnetic mechanism to evidence.
This cross-cutting project connects high-fidelity electromagnetic simulation and experimental data to reusable inference workflows. Instead of optimizing only predictive accuracy, models are evaluated for stability, uncertainty, feature dependence, and compatibility with the underlying electromagnetic mechanism.
EM modelFields and S-parameters
ValidateMeasurement / numerical consistency
RepresentPhysics-informed features
InferML, inverse modelling and uncertainty
Evidence & related work
THz sensing + classifier benchmarking
A 2026 THz sensing study benchmarks multiple supervised classifiers on features derived from simulated electromagnetic response.
Publication DOI ↗Geometry-aware EM modelling
Earlier work establishes a rigorous computational route for generating and analysing complex deformed RF structures.
Publication DOI ↗