Get Started
Home
Topics
Search
Library
Research questionHow can compact device surrogates stay physically consistent and accurate across unseen process or geometry splits with scarce data?Sparse device data make it difficult to learn behavior that remains reliable when fabrication processes or device geometries differ from those observed during training. The surrogate must capture physically consistent trends while extrapolating across these distribution shifts.
AI
Machine Learning
Neural and Evolutionary Computing
Reinforcement Learning
Research Paper
Statistical Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.A Unified Physics-Aware Quantum Machine Learning Framework across Power GaN HEMTs and Logic Nanowire FETs: Predicting Unseen Process Splits and Held-Out Geometry Combinations with Lower Error and Tighter Split-to-Split VariabilityThe evidence concerns an RL framework that searches compact parametrized quantum circuits with a graph neural network policy and PPO for Power GaN HEMTs and logic nanowire FETs. The circuits are classically simulated and evaluated with leave-one-group-out cross-validation on two device datasets, using error and split-to-split variability against six classical baselines; broader device coverage and deployment performance are not established.research paper · Sep 4, 2026
Related questions
How can surrogate models accelerate TCAD exploration of large 3D FinFET designs without losing physics across geometries?How can we estimate simulator-parameter distributions that reproduce observations without inaccurate fixed likelihood surrogates?How can storm-surge emulators generate diverse spatial scenarios for hypothetical storms without costly physics simulations?How can physics-informed neural networks handle stiff, multiscale ion-electronic transport while supporting inverse estimation?