Date & Time: 

   2026 /06 / 05  (Fri) 14:20 - 16:20

 

Location: 

   Delta Building R216, NTHU

 

Speaker: 

  祁忠勇 教授

  國立清華大學 電機資訊學院

 

Topic: 

  CVXopt-Aided AI for HSI Super-Resolution: CAUWT

 

Abstract: 

  In recent years, fusing high-resolution multispectral images (HR-MSIs) and low-resolution hyperspectral images (LR-HSIs) has become a widely used approach for hyperspectral image super-resolution (HSI-SR). The deep unfolding framework has attracted significant attention thanks to its ability to formulate the problem into a data module and a prior module. However, there are still two critical issues that hinder the performance enhancement of the existing methods: 1) Parameters in the data module are fixed (though learnable) at each iteration, i.e., lacking the adaptivity to comprehensive data; 2) The Transformer in the prior module cannot effectively capture high-frequency information. To resolve these issues, we recently proposed a Content-Adaptive Unfolding Wavelet Transformer (CAUWT) for HSI-SR, which primarily consists of four deep learning sub-networks. The CAUWT to be presented is actually a Convex Optimization (CVXopt)-Aided Artificial Intelligence (AI) approach for HSI-SR. Specifically, the parameters in the data are adaptively learned based on the reconstructed HSI at each iteration, embodied by a sub-network, termed Parameter Prediction (PP); effectively capturing the high-frequency components is achieved by the Discrete Wavelet Transform (DWT), embodied in a novel sub-network, termed Wavelet-Assisted Transformer (WAT), (which is done for the first time to the best of our knowledge); a sub-network, termed Gradient Update Module (GUM), remarkably downgrades the tremendous computational complexity. Extensive experiments on both simulated and real datasets demonstrate that CAUWT outperforms state-of-the-art methods in overall performance. Finally, we conclude the presentation with a proactive trend: Intelligent Fusion of CVXopt & AI.

 

Autobiography: 

  Chong-Yung Chi (IEEE Life Fellow, AAIA & AIIA Fellows, NAAI Member) received a B.S. degree from Tatung Institute of Technology, Taipei, Taiwan, in 1975, an M.S. degree from National Taiwan University, Taipei, Taiwan, in 1977, and a Ph.D. degree from the University of Southern California, Los Angeles, CA, USA, in 1983, all in electrical engineering. Currently, He is a Professor at National Tsing Hua University, Hsinchu, Taiwan. He has published more than 240 technical papers (with citations more than 8900 by Google-Scholar), including more than 100 journal papers (mainly in IEEE TRANSACTIONS ON SIGNAL PROCESSING), more than 140 peer-reviewed conference papers, 3 book chapters, and 2 books, including a textbook, Convex Optimization for Signal Processing and Communications: From Fundamentals to Applications, CRC Press, 2017 (which has been popularly used in a series of invited intensive short courses at 10 top-ranking universities in Mainland China since 2010 before its publication). Dr. Chi received the 2018 IEEE Signal Processing Society Best Paper Award. He has been a member of the Technical Program Committee for many IEEE-sponsored and cosponsored workshops, symposia, and conferences on signal processing and wireless communications. His research interests include signal processing for wireless communications, convex analysis and optimization for blind source separation, biomedical and hyperspectral image analysis, and, in particular, Intelligent Fusion of Convex Optimization and Artificial Intelligence.