Electro-optical Neural Networks based on Time-stretch Method
This addresses hardware acceleration for neural networks, but it is incremental as it builds on existing optical computing methods.
The paper tackled implementing neural networks using an electro-optical time-stretch method, achieving 88% accuracy on a handwriting digit recognition task with noise.
In this paper, a novel architecture of electro-optical neural networks based on the time-stretch method is proposed and numerically simulated. By stretching time-domain ultrashort pulses, multiplications of large scale weight matrices and vectors can be implemented on light and multiple-layer of feedforward neural network operations can be easily implemented with fiber loops. Via simulation, the performance of a three-layer electro-optical neural network is tested by the handwriting digit recognition task and the accuracy reaches 88% under considerable noise.