Green Learning for OFDM Signal Classification
Data-driven deep learning (DL)-based wireless systems have received widespread attention in the communications community in recent years. While DL for wireless communications is promising to improve performance and break boundaries in traditional communication architectures, many challenges such as lack of interpretability and tractability as well as large training overhead still need to be resolved in order for the marriage between DL and wireless communications to be fully embraced by the academia and even more so by the industry. To simultaneously address the issues of computational complexity and model interpretability, a suite of novel, powerful and yet very low-complexity machine learning models, termed as the green learning (GL), has been developed recently to tackle various problems in computer vision with the specific goal of achieving low carbon footprint and high logical transparency with modularized designs. The design of a GL system represents a significant deviation from all modern DL architectures. Notably, GL does not require backpropagation learning process, which contributes to substantial savings in computations while offering comparable or even better inference performance. In this talk, I will briefly introduce the fundamentals of green learning, show its performance advantages over other DL-based approaches in computer vision applications, and discuss how green learning can be applied to replace the DL block in classical DL-based OFDM signal detection.
Feng-Tsun Chien received the B.S. degree from National Tsing Hua University, Hsinchu, Taiwan, in 1995, the M.S. degree from National Taiwan University, Taiwan, in 1997, and the Ph.D. degree from the University of Southern California (USC), Los Angeles, CA, USA, in 2004, all in electrical engineering. From 2005 to 2021, he was with the Department of Electronics Engineering, National Chiao Tung University, Hsinchu, Taiwan. Since 2021, Dr. Chien has been with the Institute of Electronics and the Institute of Artificial Intelligence Innovation, National Yang Ming Chiao Tung University (NYCU), where he is currently a Professor. Dr. Chien was a Fulbright Scholar with the University of California at Los Angeles (UCLA), during the academic year 2016–2017.
Dr. Chien co-directs the Communication Electronics and Signal Processing Laboratory (CommLab) in the Institute of Electronics, NYCU. He also serves as the division head of the Data Governance in the Center of Institutional Research and Data Analytics (CIRDA). His research interests include wireless communications, signal and image processing, and machine learning. Dr. Chien is currently an Associate Editor for IEEE Transactions on Signal and Information Processing over Networks and APSIPA Transactions on Signal and Information Processing.
Dr. Feng-Tsun Chien
Institute of Electronics and the Institute of Artificial Intelligence Innovation, National Yang Ming Chiao Tung University (NYCU), Taiwan