Unifying and Merging Well-trained Deep Neural Networks for Inference Stage
|Speaker:||Dr. Yi-Ming Chan|
|Date:||2018-07-03 (Tue) 14:00 - 15:30|
|Location:||Auditorium 122 at CITI Building|
We propose a novel method to merge convolutional neural-nets for the inference stage. Given two well-trained networks that may have different architectures that handle different tasks, our method aligns the layers of the original networks and merges them into a unified model by sharing the representative codes of weights. The shared weights are further re-trained to fine-tune the performance of the merged model. The proposed method effectively produces a compact model that may run original tasks simultaneously on resource-limited devices. As it preserves the general architectures and leverages the co-used weights of well-trained networks, a substantial training overhead can be reduced to shorten the system development time. Experimental results demonstrate a satisfactory performance and validate the effectiveness of the method.
Bio:Yi-Ming Chan was born on February 27, 1978 in Taipei, Taiwan. He received the B.S. degree in mechanical engineering from National Chung Hsing University, Taichung, Taiwan, in 2000 and the M.S. degree in computer science from National Taiwan University, Taipei, Taiwan, in 2005. He received the Ph.D. degree in computer science in the Department of Computer Science and Information Engineering, College of Electrical Engineering and Computer Science, National Taiwan University, Taipei, Taiwan. His research interests include computer vision, pattern recognition, target tracking, robotics, deep learning and intelligent transportation systems.