IEEE SPS Distinguished Lecturer 講演会 (2025年11月 3会場にて開催)
IEEE SPS Distinguished LecturerのMathews Jacob教授をお招きしてIEEE SPS Distinguished Lecturer 講演会を⼤阪，横浜，⽔⼾の3会場にて開催します。奮ってご参加ください。
参加方法
対面参加の場合は，直接現地にお越しください。
大阪開催はAPSIPA DL講演と合同で⾏われます。
APSIPA DL講演は，東北⼤学 教授の栗林稔先⽣により”Ownership Verification of AI Models”と題して，同会場にて10:45-11:45に⾏われます。
横浜開催でのオンライン参加の場合は，メーリングリストに案内されているURLにアクセスください。
水戸開催は 信号処理シンポジウム内で行われます。
本講演はいずれもどなたでも無料で参加いただけますが，水戸開催につきましては本講演のほか，シンポジウムの各セッションに参加するには別途登録が必要になる旨ご承知おきください。
講演概要
- Lecturer
- Prof. Mathews Jacob (University of Virginia, USA), IEEE Distinguished Lecturer
- Biography
- Mathews Jacob is a Professor in the Department of Electrical and Computer Engineering at the University of Virginia. His research interests lie in computational medical imaging, with a focus on the intersection of physics-based modeling and machine learning. He received his Ph.D. from the Swiss Federal Institute of Technology and was a Beckman Postdoctoral Fellow at the University of Illinois at Urbana-Champaign. He is the recipient of the NSF CAREER Award, the Research Scholar Award from the American Cancer Society, the Faculty Excellence Award for Research from the University of Iowa, and three IEEE best conference paper awards. He also has served as the general chair of the IEEE International Symposium on Biomedical Imaging in 2020. He is also a Fellow of the IEEE and is a Distinguished Lecturer of the IEEE Signal Processing Society.
- (1) 大阪会場（APSIPA DL講演との合同開催）
- 日時 (Date and Time)
- 11/21 (Fri.) 9:30-10:30
- 場所 (Place)
- ⼤阪⼤学中之島センター 6F セミナー室6A，6B
〒530-0005 ⼤阪府⼤阪市北区中之島4-3-53
アクセス方法 - (2) 横浜会場（ハイブリッド開催）
- 日時 (Date and Time)
- 11/21 (Fri.) 15:00-16:00
- 場所 (Place)
- 《対面》横浜国⽴⼤学 理⼯学部講義棟(S5-7) A107教室
〒240-8501 神奈川県横浜市保⼟ケ⾕区常盤台79-5
アクセス方法
《オンライン》Google Meet *MLに記載のURLよりアクセスください - (3) 水戸会場（信号処理シンポジウム内での開催）
- 日時 (Date and Time)
- 11/25 (Tue.) 16:30-17:50
- 場所 (Place)
- ホテルレイクビュー⽔⼾
〒310-0015 茨城県⽔⼾市宮町1-6-1
アクセス方法 - 大阪/横浜会場
- Title
- Computational MR imaging in the AI era
- Abstract
- Magnetic Resonance Imaging (MRI) is undergoing a paradigm shift driven by advances in computational techniques and artificial intelligence. This talk will introduce the challenges and opportunities in the area of medical imaging. It will then explore the evolving landscape of computational MRI, where physics-based modeling, signal processing, and machine learning converge to accelerate acquisition, enhance image quality, and enable new diagnostic capabilities. It will start with supervised machine learning approaches, followed by new advances in generative artificial intelligence tools.
- 水戸会場
- Title
- Multi-scale energy (MuSE) models for imaging with guarantees
- Abstract
- Magnetic Resonance Imaging (MRI) is undergoing a paradigm shift driven by advances in computational techniques and artificial intelligence. I will introduce state of the art approaches such as unrolled methods, plug and play priors, and diffusion models that are being increasingly used in MRI. I will then introduce multiscale energy models (MuSE), which are related to diffusion models for image recovery. Unlike classical diffusion models, MuSE models explicitly model the prior and posterior distributions in image recovery tasks. This property enables one to formulate the inference as an optimization algorithm rather than computationally expensive ordinary/stochastic differential equation solvers used in the diffusion context. We introduce fast optimization algorithms with guaranteed convergence as well as introduce guarantees on robustness. The utility of this scheme to provide fast inference and uncertainty estimates will be shown in several applications, including static and dynamic MRI.