專題演講

Abstract:

Mobile/edge visual analytics will prevail in the modern AI era. Most researchers focus on deep-learning-based model compression to achieve this goal. Model compression can reduce the model size by 50-80% with slight performance degradation. Model compression relies on an existing larger model. The training cost of such a large model remains. The compression step also demands resources. I have worked on green AI since 2014, published many papers on this topic, and coined this emerging field “green learning.” Green learning demands low power consumption in both training and inference. It has attractive characteristics, such as small model sizes, fewer training samples, mathematical transparency, ease of incremental learning, etc. It can reduce the model size of its deep-learning counterpart by 95-99%. The training can be conducted from scratch. The resulting model is inherently smaller. It is ideal for mobile and edge devices. Green learning relies on signal-processing disciplines such as filter banks, linear algebra, subspace learning, probability theory, etc. Although it exploits optimization, it avoids end-to-end system optimization, a non-convex optimization problem. Instead, it adopts modularized optimization, and each optimization problem can be cast as convex optimization. In this example, I will use several examples to demonstrate the advantages of green learning in visual analytics for mobile/edge devices.

Speaker’s Biography:

Dr. C.-C. Jay Kuo received his Ph.D. from the Massachusetts Institute of Technology in 1987. He is now with the University of Southern California (USC) as the Ming Hsieh Chair Professor, a Distinguished Professor of Electrical and Computer Engineering and Computer Science, and the Director of the Media Communications Laboratory. His research interests are in visual computing and communication. He is a Fellow of AAAS, ACM, IEEE, NAI, and SPIE and an Academician of Academia Sinica. Dr. Kuo has received a few awards for his research contributions, including the 2010 Electronic Imaging Scientist of the Year Award, the 2010-11 Fulbright-Nokia Distinguished Chair in Information and Communications Technologies, the 2019 IEEE Computer Society Edward J. McCluskey Technical Achievement Award, the 2019 IEEE Signal Processing Society Claude Shannon-Harry Nyquist Technical Achievement Award, the 72nd annual Technology and Engineering Emmy Award (2020), and the 2021 IEEE Circuits and Systems Society Charles A. Desoer Technical Achievement Award. Dr. Kuo was the Editor-in-Chief of the IEEE Transactions on Information Forensics and Security (2012-2014) and the Journal of Visual Communication and Image Representation (1997-2011). He is currently the Editor-in-Chief for the APSIPA Trans. on Signal and Information Processing (2022-2025). He has guided 180 students to their Ph.D. degrees and supervised 31 postdoctoral research fellows.

Abstract: 

Reconstructing and editing visual worlds from images lies at the heart of computer graphics and computer vision. Recent advances—driven by neural rendering and learning-based techniques—have significantly improved the fidelity of 3D reconstruction. However, several challenges remain, especially under extreme lighting variation, complex material properties, and limited observations. In the first part of this talk, I will present our recent work addressing these challenges. We propose novel methods that improve robustness to lighting changes and enable accurate reconstruction in diverse scenarios. In the second part, I will discuss how we can leverage learning-based priors for editing images/videos without explicitly modeling complex visual appearances. Finally, I will conclude with emerging trends and open challenges. 

Speaker’s Biography: 

Jia-Bin Huang is a Capital One-endowed Associate Professor of Computer Science at the University of Maryland, College Park. Before coming to UMD, Huang was a research scientist at Meta Reality Labs. Before that, he was an Assistant Professor of Electrical and Computer Engineering at Virginia Tech. Huang received his Ph.D. from the University of Illinois, Urbana-Champaign (UIUC) in 2016. His research interests include 3D computer vision, generative models, and computational photography. Huang is the recipient of the Thomas & Margaret Huang Award, NSF CRII award, faculty award from Samsung, Google, 3M, Qualcomm, and a Google Research Scholar Award. 

Abstract: 

Superhuman Sports fuses technology and culture with athletics, enabling people to acquire a new, machine-integrated body and enjoy sports as “superhumans.” Advanced technologies—information processing, robotics, AR/VR, and more—can let individuals surpass conventional human capabilities and transcend the barriers that arise from physical differences among people. After introducing the concept of Superhuman Sports, the talk will share insights gained by Ryoichi Ando and colleagues through their research in human-augmentation engineering and their activities aimed at realizing an inclusive society.

Speaker’s Biography:

Ryoichi Ando received his Ph.D. from Keio University Graduate School of Media Design in 2023. He currently serves as President of AXEREAL Co., Ltd., Director of the Superhuman Sports Project, Visiting Professor at the Professional University of Information and Management for Innovation, and Researcher at Keio University. His work explores the intersection of human augmentation, sports, and pop culture. As founding director of the Superhuman Sports Society, he helped pioneer a new genre of inclusive sports, receiving the Good Design Award (2017) and international media attention, including from BBC. His doctoral project SliDeRift—an augmented wheelchair enabling drifting— was exhibited globally and inspired the creation of a new sport. He also developed SpiritOverflow, a digitally enhanced sport promoting physical diversity, deployed across five countries with support from the Japan Sports Agency. In 2018, Dr. Ando founded AXEREAL to develop tech-driven sports, integrating robotics, sensing, and tele-existence technologies in collaboration with universities and industry.