Research
My research lies at the intersection of scientific machine learning, computational electromagnetics, and AI for physics. I develop physics-informed and data-driven numerical approaches for electromagnetic forward and inverse problems, with applications in power systems, biomedical engineering, and geophysical imaging.
Scientific Machine Learning for PDEs
I investigate physics-informed neural networks (PINNs), variational and weak-form formulations, and neural operators for solving and accelerating electromagnetic partial differential equations.
Selected research:
- INI-VPINN: Variational PINNs with implicit Neumann and interface handling in multi-material domains. Paper · Code
- STAR-PINN: Stacked adaptive residual PINNs for nonlinear magnetic diffusion. Paper · Code
- Hybrid BEM–PINN: Boundary element methods combined with physics-informed learning for electromagnetic problems. Paper · Code
- Neural operators: Physics-informed DeepONet surrogate modeling of parametric electromagnetic devices; ongoing work on physics-informed Fourier neural operators. Paper
AI for Power and Energy Systems
Within the FELINES research project, I developed machine-learning approaches to lightning localization, peak-current estimation, and lightning-induced overvoltage prediction, supporting computational tools for power-system protection.
Selected research:
AI for Biomedical Electromagnetics
My research includes generative learning and inverse modeling for patient-specific transcranial magnetic stimulation (TMS), as well as deep-learning surrogate models for rapid assessment of specific absorption rate (SAR) in electromagnetic devices.
Selected research:
- STEM-DEEP: Generative electric-field modeling and inverse coil-placement optimization for TMS. Research program
- SAR surrogate modeling: Rapid prediction of SAR hotspots. Paper
Geophysical Inversion and Scientific Computing
My work on AIGEO focuses on physics-informed and data-driven approaches to geophysical electromagnetic inverse problems, including electrical resistivity tomography (ERT). My scientific-computing background also includes GPU-accelerated FDTD simulation for wave propagation in plasma.
Selected research:
- AIGEO: AI for electromagnetic geophysical inversion (ongoing research)
- GPU-accelerated FDTD: Electromagnetic wave propagation in plasma. Preprint
For research announcements, visit my blog. For my full academic record, see the CV.