AI vision for semiconductor manufacturing
Virtual metrology and wafer map classification for intelligent manufacturing.
Developed computer vision and machine learning algorithms for semiconductor manufacturing. Projects included virtual metrology from structured production data and wafer map pattern classification under limited labels, annotation noise, multiple labels, and open-set category issues.
Methods included image interpolation, super-resolution reconstruction, attention mechanisms, autoencoders, variational autoencoders, transfer learning, OpenSet loss, and Center loss.