Singapore University of Technology and Design (SUTD), A*STAR, and the National University of Singapore (NUS) have proposed a self-supervised attention-based virtual metrology framework, GREP, that predicts CMP MRR through global surrogate target regression. The framework first learns temporal representations from unlabeled sensor signals, followed by supervised fine-tuning, and is compared with a feature-engineering supervised baseline under three conditions: limited labeled data, sensor noise, and temporal process drift.
Source:Semiconductor Engineering · Read original
Leave a Reply