跳到主要內容區塊
:::

專題演講|Transformation-identical CNN(TI-CNN) and Geared Rotation-identical CNN (GRI-CNN) and Their Variants for Orientation-Independent Pattern Classification.(2019年10月9日)

◎活動地點:國立成功大學資訊工程系新館65104教室
◎活動日期:108年10月09日(三)
◎活動時間:10am-12pm
◎演講題目:Transformation-identical CNN(TI-CNN) and Geared Rotation-identical CNN (GRI-CNN) and Their Variants for Orientation-Independent Pattern Classification.

◎講者:婁世鐘教授 (Arlington Innovation Center, Physics Department, Virginia Tech, Arlington, VA / Radiology and Oncology Department, Georgetown University, Washington, DC)

◎講者簡介:
Dr. Lo received his Ph.D. in Medical Physics from UCLA in 1986. In 1987 he joined Center for Imaging Sciences and Information Systems (ISIS) at Radiology and Oncology Departments of Georgetown University. In 2014, he moved to Arlington Innovation Center, Virginia Tech with his research team. Dr. Lo was one of earliest researchers in deep learning convolutional neural networks (CNN) since early 1990s. He also invented wavelet CNN (1995) and circular path CNN (1998). In addition, he has performed research in optimization of wavelets and filter bank unification theory through CNN. In 2018, he invented 3 versions of transformation-identical CNN (TI-CNN) and also showed that rotation-identical CNN (GRI-CNN) with a small angle step can be constructed as an isotropic CNN in a quantitative manner. Recently, he further developed multi-perspective CNN (MP-CNN) for general image pattern recognition. Since 1990s, he has collaborated with his industrial collaborators and have developed several US-FDA cleared medical AI systems including (a) rib-suppressed chest X-ray + automatic lung nodule/cancer and disease detection AI system; (b) blood-vessel removed thoracic CT + automatic lung nodule/cancer and disease detection AI system.

報名請點我