Material computing based on physicochemical dynamics toward neuromorphic hardware implementation
Yuki Usami
Kyushu Institute of Technology, Japan
DEIB - Conference Room "E. Gatti" (Building 20)
On Line via Webex
June 13th, 2022
11.00 am
Contacts:
Daniele Ielmini
Research Line:
Electron devices
Kyushu Institute of Technology, Japan
DEIB - Conference Room "E. Gatti" (Building 20)
On Line via Webex
June 13th, 2022
11.00 am
Contacts:
Daniele Ielmini
Research Line:
Electron devices
Abstract
On June 13th, 2022 at 11.00 am, Yuki Usami, Assistant Professor at Department of Human Intelligence Systems, Graduate school of Life Science and Systems Engineering, Kyushu Institute of Technology in Japan, will hold a seminar on "Material computing based on physicochemical dynamics toward neuromorphic hardware implementation" in DEIB Conference Room and in live streaming via Webex.
Mimicking brain function by nanomaterials is an innovative approach for an effective information processing system. Reservoir computing (RC), a kind of artificial neural network, is an attractive learning method with low power consumption and fast calculation because weight update was carried out at only readout between reservoir layer and output layer. Owing to this advantage, reservoir layer can utilize even dynamic physical behaviors. Robust nonlinearity and memory properties are needed to use the material as an RC device. To achieve the requirement, we propose to utilize a highly conductive material network containing nonlinear-conduction regions.
Further details in attachment. The event will be held online by Webex.
Mimicking brain function by nanomaterials is an innovative approach for an effective information processing system. Reservoir computing (RC), a kind of artificial neural network, is an attractive learning method with low power consumption and fast calculation because weight update was carried out at only readout between reservoir layer and output layer. Owing to this advantage, reservoir layer can utilize even dynamic physical behaviors. Robust nonlinearity and memory properties are needed to use the material as an RC device. To achieve the requirement, we propose to utilize a highly conductive material network containing nonlinear-conduction regions.
Further details in attachment. The event will be held online by Webex.
Short Bio
Yuki Usami is an Assistant Professor at Department of Human Intelligence Systems, Graduate school of Life Science and Systems Engineering, Kyushu Institute of Technology, Japan. He is also a member of Research Center for Neuromorphic AI Hardware at Kyushu Institute of Technology. He received Ph.D. in Science from Osaka University at 2020. His research is focuses on the development of unconventional computing system composed of nanomaterials, especially reservoir computing.
Education:
2017-2020 Ph.D (Science), Osaka University, Japan
2015-2017 MSc., Osaka University, Japan
Work Experience:
2020-present Assistant Professor, Kyushu Institute of Technology, Japan.
