“电子信息名家讲坛” 是电子工程系主办的系列活动，旨在建立世界一流的电子信息领域学术交流平台，拓展广大师生的学术视野。
本期讲坛有幸邀请David J. Brady作题为
The history and the future of optical neural networks的报告。
讲座时间、地点
10月9日14:00-15:30
地点：清华大学电子工程馆9-208
演讲主题
The history and the future of optical neural networks
光神经网络：回溯与展望
演讲内容简介
Optical neural networks were developed in the 1980s as associative memories implementing Hopfield-model networks. While these systems were not immediately technologically successful, they included innovations and insights that remain more advanced than some modern diffractive optical systems. The success of modern artificial networks has derived from compact kernels and small-scale interconnections that are compatible with electronic processors. This success suggests, however, that a return to large-scale associate memories built on optical components may yield further progress. We consider particular opportunities in the context of super cameras.
演讲嘉宾介绍
David J. Brady is the J. W. and H. M. Goodman Endowed Professor of Optical Sciences at the University of Arizona, where he leads the Camera Lab. Brady is a Fellow of IEEE, Optica and SPIE and the recipient of the SPIE Denis Gabor Award and the Optica Leith Medal. This talk is built on material from Brady's Ph. D. Thesis, "Photorefractive Volume Holography in Artificial Neural Networks." His Ph. D. advisor was Professor Demetri Psaltis and Professor John Hopfield was a member of his committee.
供稿｜信息认知与智能系统研究所
编辑｜陶旋姿
审核｜汪 玉 沈 渊 李冬梅
