Abstract
We have designed, fabricated, and tested a set of scalable compact optical elements for use in the compact optoelectronic integrated neural (COIN) coprocessor. The COIN coprocessor is an implementation of a feed-forward neural network that uses optical interconnects to transmit information from one layer of neurons to the next. Optical interconnection elements based on Fresnel-zone-plate concepts have been fabricated using standard lithographic processes, thereby allowing large arrays to be fabricated with a single exposure. This paper discusses the design trade-offs that must be considered for effective use of such elements in a COIN coprocessor, and the fabrication and testing of a set of such optical interconnection elements. It was found that the behavior of the fabricated zone plates correlated well with the design simulation, and that this type of optical element is indeed suitable for use in the COIN coprocessor.
© 2017 Optical Society of America
Full Article | PDF ArticleMore Like This
Huarong Gu and Cardinal Warde
Appl. Opt. 55(30) 8576-8581 (2016)
Kashiko Kodate, Eriko Tokunaga, Yukiko Tatuno, JingLi Chen, and Takeshi Kamiya
Appl. Opt. 29(34) 5115-5119 (1990)
Shulang Lin, Huarong Gu, and Cardinal Warde
Appl. Opt. 57(29) 8664-8672 (2018)