Machine learning-based lithography feature selection for SMO and Model Calibration

Next-generation lithography models and optimized masks require careful selection of features for generating accurate models or sources. Thus, choosing representative features is crucial for building better models and achieving optimized sources. The goal of this study is to identify representative features for both Source Mask Optimization (SMO) and model calibration. To accomplish this, the study will employ both supervised and unsupervised machine learning techniques.