All research areas
AI-Assisted Semiconductor Device Modeling
Machine-learning methods for fast, accurate modeling of active and passive semiconductor devices when measurement data are limited.
Accurate device models are essential for designing integrated circuits at millimeter-wave and sub-terahertz frequencies. We investigate how machine learning can complement measurement-driven characterization and conventional compact modeling for semiconductor devices.
Research directions
- Machine-learning-assisted models for active and passive devices
- Neural-network models for transistor behavior
- Learning from small and measurement-limited datasets
- Data augmentation for device characterization
- Fast surrogate models for circuit design and optimization
This research combines high-frequency measurement expertise with data-driven modeling. It supports faster design-space exploration while preserving the accuracy needed for analog/RF and sub-terahertz circuit development.