Large-scale spatiotemporal photonic reservoir computer for image classification
This work addresses efficient hardware implementation of neural networks for image classification, but it is incremental as it applies existing reservoir computing methods to photonic systems.
The authors tackled image classification of handwritten digits using a scalable photonic reservoir computing architecture, achieving a 1% error rate on MNIST with a feedforward network of 16,384 nodes, which is state-of-the-art for experimental implementations.
We propose a scalable photonic architecture for implementation of feedforward and recurrent neural networks to perform the classification of handwritten digits from the MNIST database. Our experiment exploits off-the-shelf optical and electronic components to currently achieve a network size of 16,384 nodes. Both network types are designed within the the reservoir computing paradigm with randomly weighted input and hidden layers. Using various feature extraction techniques (e.g. histograms of oriented gradients, zoning, Gabor filters) and a simple training procedure consisting of linear regression and winner-takes-all decision strategy, we demonstrate numerically and experimentally that a feedforward network allows for classification error rate of 1%, which is at the state-of-the-art for experimental implementations and remains competitive with more advanced algorithmic approaches. We also investigate recurrent networks in numerical simulations by explicitly activating the temporal dynamics, and predict a performance improvement over the feedforward configuration.