Subresolution Assist Feature Insertion by Variational Adversarial Active Learning and Clustering with Data Point Retrieval
Hosted in Virtual Platform
Physical Design and Verification, Lithography and DFM
DescriptionSubresolution assist feature (SRAF) insertion is one of resolution enhancement techniques to improve the target pattern printability. In advanced lithography, we have a huge solution space but few labeled training samples. Therefore, we present a novel variational adversarial active learning framework, not relying on the model uncertainty to select informative samples. Second, we propose a region-based concentric circle area sampling representation to avoid information loss during feature extraction. Third, we determine the final SRAF placement by clustering using retrieved data points. Experimental results demonstrate, compared with state-of-the-art works, our framework uses 40% training samples and improves PV band and EPE.