Title: Energy (Battery)-Efficient Kernel-Based SVM Classification
Speaker： Professor S.Y. Kung, Princeton University
Time： May 28, 2010 10am-11am
Venue： FIT 1-515
Contact： Shiqiang YANG, FIT 3-518 62771978
Abstract:
Green computing with a major emphasis placed on energy efficiency is vital for mobile multimedia devices or implanted medical systems, which disallow frequent replacement of the power supply. On the other h, kernel-based methods have recently dominated the world of machine learning. This talk proposes a beneficiary marriage between these two important fronts of technologies. We advocate a vertically integrated approach to green computing, combining both perspectives of the circuit-device system (machine learning) levels. Our goal is to expedite the retrieval processing time /or save the processing energy for on-line prediction/detection. First, we demonstrate that the knowledge on hardware device noise may be used to train an error-resilient classifier. Second, an SNR-based feature ion tool will be proposed as dimension reduction tool, which has advantage over the traditional PCA (principal component analysis). Third, it will be shown that the complexity hinges upon two factors (1) the number of basis functions induced by the kernel (2) the support vectors required to form the decision rule.
Bio:
Professor S.Y. Kung received his Ph.D. Degree in Electrical Engineering from Stanford University in 1977. Since 1987, he has been a Professor of Electrical Engineering at the Princeton University. His research interests include VLSI array processors, system modeling identification, neural networks, wireless communication, sensor array processing, multimedia signal processing, bioinformatic data mining biometric authentication. Professor Kung is a Fellow of IEEE since 1988. He served as a Member of the Board of Governors of the IEEE Signal Processing Society (1989-1991). He was a founding member of several Technical Committees (TC) of the IEEE Signal Processing Society. Since 1990, he has been the Editor-In-Chief of the Journal of VLSI Signal Processing Systems.
