Communications of the ACM
World's First Silicon Photonic Neural Network nveiled
Credit: AI Labs
Alexander Tait and colleagues at Princeton University say they have built the world's first integrated silicon photonic neuromorphic chip, a development that could lead to superfast optical computers based on neural networks. The laser output from their optical device is mathematically equivalent to a device known as a continuous-time recurrent neural network, which suggests programming tools for such networks could be applied to larger silicon photonic neural networks.
The device can immediately exploit the vast range of programming nous that has been gathered for these kinds of neural networks. The team demonstrated how this can be done using a network consisting of 49 photonic nodes, and the results show just how fast photonic neural nets can be. "The effective hardware acceleration factor of the photonic neural network is estimated to be 1,960× in this task," according to Tait and colleagues.
That is a speed up of three orders of magnitude, which opens the doors to a new industry that could bring optical computing into the mainstream for the first time. The result could be ultrafast information processing for radio, control, and scientific computing.
From Technology Review
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA
No entries found