Speaker: 전동석 조교수(융합과학기술대학원)
Date & Time: 2018.11. 27.(화), 17:00
Where: 융대원 D-122호
Deep learning algorithms have gathered serious attention in the last few years due to their outstanding performance even comparable to that of a human. Their application areas are fast expanding from computer vision and speech recognition to multi-modal understanding. However, due to the highly constrained power budget of battery-powered systems, and unavoidable leakage and enlarged PVT variability of sub-nanometer process, we are still far from realizing state-of-the-art machine learning algorithms on mobile platforms.
Neuromorphic computing seeks to mimic the mechanism of human brain precisely by sticking to what neuroscientists have discovered in the brain. Spiking neural network has been studied as a strong candidate due to its biological plausibility. The algorithm was proposed decades ago, but its performance still remains inferior to that of other deep learning algorithms.
In this talk, various hardware designs aimed at realizing those algorithms on mobile platforms will be discussed in detail. Since a large network is a necessity in real-world applications, reducing hardware cost and power consumption is a key to successful hardware implementation. Examples of hardware accelerators for deep learning algorithms as well as neuromorphic computing will be presented along with their advantages and limitations. Finally, some recent works from our research group will be introduced.
Dongsuk Jeon received a B.S degree in electrical engineering from Seoul National University, South Korea, in 2009 and a Ph.D. degree in electrical engineering at the University of Michigan, Ann Arbor in 2014. From 2014 to 2015, he was a postdoctoral associate at Massachusetts Institute of Technology. He is currently an assistant professor of Graduate School of Convergence Science and Technology at Seoul National University. His research interests include energy-efficient signal processing, low power circuit and SoC for mobile applications.
초청자 : 융합과학부 지능형융합시스템전공 박재흥 교수 (firstname.lastname@example.org)