주제주제
Machine learning (ML) and deep learning (DL) have been pivotal in solving complex problems across physics, chemistry, and engineering. In semiconductor process and device design, the integration of domain-specific knowledge into ML/DL models is essential due to the limited availability of data, bridging the gap between theoretical simulations and practical applications. ML and DL have driven significant advancements in semiconductor manufacturing. Applications include real-time prediction of doping profiles and current-voltage characteristics in MOSFETs, full-chip stress analysis using hybrid frameworks, and compact modeling for scalable device designs. These innovations not only enhance alignment with experimental results but also optimize device performance under constrained conditions. Strategies such as transfer learning, which employs pre-trained models to fine-tune tasks with limited data, further address challenges posed by data scarcity, expanding the scope of ML/DL applications in semiconductor engineering
