Wen-Hsiao Peng Dept. of Computer Science, National Chiao Tung University, Taiwan
Application of reinforcement learning to video encoder control and video prediction
Reinforcement learning (RL) has proven effective for solving decision making problems. While it achieves great success in gaming, robotics, etc., its application to modern video codecs has yet to be seen. In this talk, I shall present three attempts to introduce RL to HEVC/H.265 video encoder control and video prediction. The first is to determine a quantization parameter value for every coding tree unit in a video frame, the objective being to minimize the frame-level distortion subject to a rate constraint. The second extends the very same idea to frame-level bit allocation within a group of video frames. The last switch gears to address the task of video frame extrapolation, for which we show how a neural network-based agent can be combined with the conventional video prediction technique within the RL framework to achieve the state-of-the-art performance with better generalization capability. The talk assumes no knowledge of reinforcement learning and video coding, and is aimed at showcasing the potential of RL in the area of visual signal compression and processing.
Dr. Wen-Hsiao Peng (M’09-SM’13) received his Ph.D. degree from National Chiao Tung University (NCTU), Taiwan in 2005. He was with the Intel Microprocessor Research Laboratory, USA from 2000 to 2001, where he was involved in the development of ISO/IEC MPEG-4 fine granularity scalability. Since 2003, he has actively participated in the ISO/IEC and ITU-T video coding standardization process and contributed to the development of SVC, HEVC and SCC standards. He is currently a Professor with the Computer Science Department, NCTU, and was a Visiting Scholar with the IBM Thomas J. Watson Research Center, USA, from 2015 to 2016. He has authored over 70 journal/conference papers and over 60 ISO/IEC and ITU-T standards contributions. His research interests include video/image coding, deep/machine learning, multimedia analytics, and computer vision. Dr. Peng is Chair of the Visual Signal Processing and Communications (VSPC) Technical Committee of the IEEE Circuits and Systems (CAS) Society. He was Technical Program Co-chair for 2011 IEEE VCIP, 2017 IEEE ISPACS, and 2018 APSIPA ASC; Publication Chair for 2019 IEEE ICIP; Area Chair for IEEE ICME and VCIP; and Review Committee Member for IEEE ISCAS. He served as Associate Editor-in-Chief for Digital Communications/Lead Guest Editor/Guest Editor/SEB Member for IEEE JETCAS, Associate Editor for IEEE TCSVT, and Guest Editor for IEEE TCAS-II. He was Distinguished Lecturer of APSIPA.