Hyunsoo Yang
Biography : Hyunsoo Yang is a Professor in the Department of Electrical and Computer Engineering, National University of Singapore (NUS), working on various magnetic materials and devices for spintronics applications. He worked at C&S technology and Intelligent Fiber Optic Systems, California.
He received his Doctorate from Stanford University. From 2004-2007, he was at IBM Almaden Research Center. He has authored more than 270 journal articles, given 200 invited presentations, and 20 patents.
Major Achievement : He was a recipient of the Outstanding Dissertation Award from the American Physical Society (GMAG), IEEE Magnetics Society Distinguished Lecturer, Minister of Science ICT award, Mid-Career Award of the IEEE Magnetics Society, AAIA Fellow, and IEEE Fellow.
Lecture Summary : Magnetic tunnel junction (MTJ)-based true random number generators (TRNGs) provide distinct advantages over conventional CMOS implementations. We demonstrate an MTJ-based TRNG, characterize the entropy of its raw output, and extract random bits with provable security. Beyond randomness, MTJs also enable unconventional computing.
We realize a proof-of-concept computing system that employs MTJs to implement sigmoidal activation functions, enabling the training of a neural network for speech separation. Furthermore, we present an experimental Ising computer comprising 250 MTJs, which successfully solves representative NP-hard problems.
In parallel, we integrate electrically connected spin-torque oscillators (STOs) to demonstrate a battery-free energy-harvesting system, converting wireless RF energy into usable power for electronic devices such as LEDs. We also highlight emerging opportunities for spin-orbit torque devices based on 2D materials, and discuss magnon-mediated spin torques with antiferromagnets as a pathway to reducing Joule heating and energy dissipation.