Technology That Matters: Strategies to Precisely Control Synaptic Weights for Neuromorphic Computing Arrays | SGInnovate


Online Event

Technology That Matters: Strategies to Precisely Control Synaptic Weights for Neuromorphic Computing Arrays

Presented by SGInnovate and SMART

Our lives have suddenly been forced into the core of the digital world this year. We are still able to work and socialise globally because of the advancement in technology. Despite the situation we are in today, scientists will continue to advance the technology to aid the human race.

As part of the Technology That Matters series, SMART and SGInnovate have invited MIT Professors to share their technologies. There will be fireside chats and panel discussions! Don’t miss this opportunity to interact with MIT Professors.
Neuromorphic computing has recently emerged as a non-Von Neuman computing method due to its analogue switching ability to represent multiple synaptic weights by varying conductance in the vertical filaments formed in the switching medium. As such, resistive memory is now considered an artificial synapse for the suitable neuromorphic hardware platform.

Conventional resistive memories typically utilise a defective amorphous solid as a switching medium for defect mediated formation of conducting filaments. However, the imperfection of the switching medium also causes stochastic filament formation leading to spatial and temporal variation of the devices. These variations of artificial synapses have prevented the community from obtaining large scale artificial neural networks.

Join us as Prof Kim Jeehwan presents on a new type of resistive memory that can increase the precision in confining the conducting paths such that uniform artificial synapses can be obtained, leading to a 100% yield of 32 x 32 arrays with great programmability. He will also share about the various material strategies to control the ionic conduction paths, which will result in low temporal/spatial variation, linear synaptic weight update, great endurance, and long retention time. Having had actual crossbar arrays manufactured, Prof Kim will also present on the properties and programmability of his team's ANN arrays.


Discover more Technology That Matters here!

Date: 28 October 2020, Wednesday
Time: 9:00am – 10:00am (Singapore Time / UTC+8)


9:00am – 9:05am: Opening Remarks by SGInnovate and SMART
9:05am – 9:35am: Presentation on Strategies to Precisely Control Synaptic Weights for Neuromorphic Computing Arrays by Prof Kim Jeehwan, MIT Department of Mechanical Engineering
9:35am – 10:00am: Fireside Chat and Q&A

Speaker’s Profile:

Prof Kim Jeehwan, MIT Department of Mechanical Engineering

Professor Jeehwan Kim is an Associate Professor of Massachusetts Institute of Technology (MIT) in the Mechanical Engineering, and Materials Science and Engineering. He is also a Principal Investigator in the Research Laboratory of Electronics at MIT. Prof Kim's group focuses on innovation in nanotechnology for next-generation computing and electronics.

Before joining MIT in 2015, he was a Research Staff Member at IBM T J Watson Research Center in Yorktown Heights, New York for seven years. In 2012, he was appointed a "Master Inventor" of IBM in recognition of his active intellectual property generation and commercialisation of his research.

Prof Kim is a recipient of DARPA Young Faculty Award, IBM Faculty Award, LAM Research Foundation Award, 20 IBM high-value invention achievement awards. He is an inventor of 200 issued/pending US patents and an author of 50 articles in various journals. He received his Bachelor of Science from Hongik University, his Masters of Science from Seoul National University, and his PhD from UCLA in 2008, all of them in Materials Science.

Topics: Artificial Intelligence / Deep Learning / Machine Learning / Robotics