The MIT research team and collaborative researchers proposed in 2016 to replace electrons with photons as the theoretical basis of computing chip architecture, also called the programmable nanophotonic processor by the research team.
In June of 2017, a paper on the programmable nanophotonic processor was proposed by the team and published on Nature Photonics. Shen Yichen, born in Hangzhou and the first author and corresponding author of this paper, is now the co-founder and CEO of Lightelligence.
Under his lead, Lightelligence is going all out to research and develop optical chip related technologies, including chip design, core algorithms, transmission, etc., to create an integrated optical computational ecology.
Lightelligence is trying to revolutionize computational ecology. As a matter of fact, photonic chips might be the most appropriate hardware architecture for AI computation in the future due to the fact that properties of light are cut out for linear computation, which is the most significant part of AI computation and includes high dimensional parallel computation.
Shen Yichen emphasized that the photonic chip of Lightelligence has currently reached lab presentation status. Designs of algorithms, buses, and storage are also underway. Lightelligence is constantly promoting the combination of algorithms with hardware and applying the corresponding chip commands and further compiling programs under frames like TensorFlow and Caffe.
Now, the Lightelligence team is striving to improve photonic computation related ecology, but the industry has high expectations for performance computation and better neural networks computing architecture. We believe when Lightelligence’s photonic computing architecture is completed, it will accelerate the revolution of the whole AI computational ecology.