In the last decades, the problems of learning decision making strategy and optimizing the resource allocation process in video coding systems have received a lot of attention. However, these two problems are still challenging due to the limited theoretical analysis tools.
Machine learning is a hotspot and widely applied in artificial intelligence, pattern recognition and signal processing, since it learns from lots of information which we probably could name it as big data in today’s terminology. With this property, researchers attempted to apply this machine learning techniques to solve the decision making problem in video coding system for better performances. Besides, bargaining game theory has been proved to be a powerful technique for addressing the limited resource allocation problem among multiple players in the collaborative systems. Resource allocation problem in video coding system can be modeled as a bargaining problem, we attempt to investigate the optimal resource allocation strategy based on bargaining game theory.
In this talk, an overview of our recent research of learning and bargaining based video coding optimization will be presented. First, an overview on traditional video coding framework will be given and the fundamental optimization problems in video coding systems will also be discussed. Second, we will present the quad-tree coding unit (CU) depth decision process in High Efficiency Video Coding (HEVC) is modeled as a three-level of hierarchical binary decision problem. Then the flexible CU depth decision structure for each decision level are proposed to learnt about a model which will achieve better performances between the coding complexity and Rate Distortion (RD) performance. Third, we will talk about modeling the inner-layer bit allocation processes of spatial scalable video coding as bargaining problems. Then the bargaining game theoretic based approaches are proposed to solve the one-pass rate control optimization problems. Experiment results will be also presented to demonstrate the performance of our proposed approaches. Finally, we will discuss the potential optimization strategy for video coding system.
Sam Kwong received the B.Sc. degree from the State University of New York at Buffalo, Buffalo, NY, in 1983, the M.A.Sc. degree in electrical engineering from the University of Waterloo, Waterloo, ON, Canada, in 1985, and the Ph.D. degree from the Fernuniversität Hagen, Germany, in 1996. From 1985 to 1987, he was a Diagnostic Engineer with Control Data Canada, where he designed the diagnostic software to detect the manufacture faults of the VLSI chips in the Cyber 430 machine. He later joined the Bell Northern Research Canada as a Member of Scientific Staff, where he worked on both the DMS-100 voice network and the DPN-100 data network project. In 1990, he joined the City University of Hong Kong as a Lecturer in the Department of Electronic Engineering. He is currently a Professor in the Department of Computer Science. He was responsible of the software design of the first handheld GSM mobile phone consultancy project in which it was one of the largest consultancy projects at the City University of Hong Kong in 1996. He coauthored three research books on genetic algorithms, eight book chapters, and over 200 technical papers. His book entitled “Genetic Algorithm for Control and Signal processing” published by Springer, London, was awarded as the Best Seller in 1997. He has been a consultant to several companies in telecommunications.
Prof. Kwong was awarded the Best Paper Award for his paper entitled “Multiobjective Optimization of Radio-to-Fiber Repeater Placement Using a Jumping Gene Algorithm” at the IEEE International Conference on Industrial Technology (ICIT’05), Hong Kong, in 2005. He also received the Best Presentation Award for the paper “Maximum Model Distance Discriminative Training for Text-Independent Speaker Verification” at the IEEE IECON 2004, Busan, Korea. He was the Invited Speaker at the IEEE 2001 ISIE Workshop in Busan, and the International Conference on Control, Automation and System held in Cheju, Korea. In addition, he received the Best Paper Award at the 1999 BioInformatics Workshop, Tokyo, for the paper entitled “A Compression Algorithm for DNA Sequences and Its Application in Genome Comparison” in recognition of his outstanding contribution to the conference.
Currently, he is the Associate Editor for the IEEE Transactions on Industrial Informatics, the IEEE Transactions on Industrial Electronics, Journal of Information Science. He also was associate editor of the Journal of Real Time Systems, from 2000 to 2010. He was also the Guest Editor of the IEEE transactions on Industrial Electronics, Feb. 2000.
Prof. Kwong has been heavily involved in organizing conferences and served closely to many IEEE Conferences and society activities. He is currently the Vice President of Conferences and Meeting of IEEE Systems, Man and Cybernetics Society. He is also a fellow of IEEE for his contributions on Optimization techniques for Cybernetics and Vide Coding.
In this big data era, large volume of data is being produced at unprecedented and explosive scale in a broad range of application areas. Analytics on such big data deliver amazing value and can drive nearly every aspect of our life. May real-world data are linked in various forms like social networks, mobile networks and Internet of Things, while big data comprised of rich and interesting information sources like sensing data, moving trajectory, social media and social relationship networks, has been formed at the same time. New opportunities arise through mining of such connected big data, which however pose many challenges, too. In this talk, I will describe some grand opportunities and key challenges therein. Some recent developments in different domains like mobile and social networks, healthcare, environmental sensing, e-commerce, etc, will be introduced first. Furthermore, some key challenging issues will be investigated in-depth through various aspects, covering data quality and veracity, key features discovery, learning methodologies, post processing and incorporation of domain knowledge. Finally, some emerging research topics and potential opportunities underlying this topic will be addressed accordingly.
Vincent S. Tseng is currently a Professor at Department of Computer Science in National Chiao Tung University, Taiwan. He has also been the chair for IEEE CIS Tainan Chapter since 2013. He served as the president of Taiwanese Association for Artificial Intelligence during 2011-2012 and acted as the director for Institute of Medical Informatics of NCKU (National Cheng Kung University) during 2008 and 2011. During 2004 and 2007, he had also served as the director for Informatics Center in Medical Center of NCKU. Dr. Tseng received his Ph.D. degree with major in computer science from National Chiao Tung University, Taiwan, in 1997. After that, he joined Computer Science Division of University of California at Berkeley as a postdoctoral research fellow during 1998-1999. He has a wide variety of research interests covering data mining, big data, biomedical informatics, mobile and Web technologies. He has published more than 300 research papers in referred journals and international conferences as well as 15 patents (held and filed). He has been on the editorial board of a number of journals including IEEE Transactions on Knowledge and Data Engineering, ACM Transactions on Knowledge Discovery from Data, IEEE Journal of Biomedical and Health Informatics, International Journal of Data Mining and Bioinformatics, etc. He has also served as chairs/program committee members for a number of premier international conferences related to data mining and intelligent computing, including KDD, ICDM, SDM, PAKDD, ICDE, CIKM, IJCAI, etc. He is also the recipient of 2014 K. T. Li Breakthrough Award.