The mission of MCL is comprised of the following two aspects:
The education of next-generation technical leaders
Professor Kuo leads MCL with two mottos: “working hard” and “being a team player.” Lab members are trained to conduct high quality research with integrity while cultivating a life-long learning habit. This consists of being open to diversified research areas, willing to communicate, and appreciative of other members’ value in a team.
The pursuit of leading and novel technologies
During the last two decades, MCL has helped develop several world-leading technologies in media analysis, coding, and transmission. Examples include: wavelet-based texture analysis, fast motion search for video coding based on spatial-temporal prediction, rate-distortion modeling for compressed video, synchronization techniques for OFDMA, and design of integrated media compression/encryption schemes. MCL will be devoted to the visual big data challenge in the next decade.
Professor Kuo hopes that the R&D experience at the Media Communications Lab will be a memorable and motivating life experience for of all its members.
Professor Kuo founded the Media Communications Lab at USC in January of 1989. In the first year, it had only three PhD students – Takang Ku, Kyoung Mu Lee and Tianhorng Chang – who conducted research on fast Toepliz solvers, shape from shading, and wavelet-based texture analysis, respectively. Over time, the lab gradually expanded and worked on variety of problems in wavelet applications and video coding. Within the topic of wavelets, the group published a series of papers on texture analysis, curve description, fractal analysis, multimedia watermarking, and environmental sound analysis. In the realm of video coding, MCL contributed to international standards such as JPEG2000, MPEG- 4, MPEG-7, H.263++. Additionally, many alumni were and are active in H.264/AVC, H.264/SVC, MVC, 3DVC, and HEVC standardization activities.
Group members have also worked in several other areas, such as wireless communications and networking, biomedical imaging, bioinformatics, multimedia forensics, and security. In 2010, MCL re-adjusted its research focus in response to the developing “big data” challenges. with the goal of efficient and robust large scale image/video data management.
Our R&D Outlook
Substantiality has become a main theme of science and technology in recent years. As civilization continues to develop, humans need be conscious in keeping the environment clean for future generations. As scientists and engineers of the 21st century, it is our destiny to keep green technologies as one of the top priorities. In the area of artificial intelligence and machine learning, it is urgent to explore a novel green learning technology, which is competitive with deep learning in performance yet with significantly lower power consumption in both training and inference.
Professor Kuo has been devoted to this subject since 2015. A sequence of papers on green learning systems has been published. Examples include: PixelHop, PointHop, FaceHop, GraphHop, GenHop, etc. These solutions have common characteristics, including low power consumption, small model sizes, weak supervision and scalability. The underlying principle of MCL’s green learning solutions is successive subspace learning (SSL).
Green learning will be the central R&D focus of MCL in the next decade. MCL members will push the envelope of green learning and develop effective green solutions for natural language processing, knowledge understanding, computer vision, joint audio-visual processing, and 3D data processing.