Guangwei Cong

AIST, Japan

Optical Streaming Processor Utilizing Integrated Photonic Circuits

Abstract

Optical computing, which processes information using optical phenomena, was explored decades ago for achieving low-latency, low-power, and extensively parallel computing. Due to fundamental limitations, it eventually failed to compete with CMOS technology. However, in recent years, the surge in computational demand driven by generative AI, along with the slowdown of Moore’s law, has reignited interest in optical computing. Advances in integrated photonics and photonics-electronics integration have significantly reduced previous limitations, making photonic systems promising candidates for future AI workloads. This talk gives a short review of the progress and challenges in this field and presents our works on developing photonic streaming processors for machine learning applications.

Biography

Guangwei Cong is currently the leader of Optical Computing Team at Photonics-Electronics Integration Research Center in the National Institute of Advanced Industrial Science and Technology (AIST), Japan. His current research focuses on the development of optical computing systems using silicon photonic circuits, including photonic neural networks, electronics-photonics hybrid computing, and enabling photonic devices. He currently also serves as the principal investigator of a project funded by the Japan Science and Technology Agency (JST) to develop photonic streaming processors. Cong received his Ph.D. degree from the Institute of Semiconductors, Chinese Academy of Sciences, in 2006 and in the same year, he joined AIST. He is a member of JSAP, IEEE, and Optica.

Download