Backpropagation Boost For Deep Learning In Photonic Neural Networks

The team at the Politecnico di Milano (Polimi) has developed a three-layer, four-port silicon photonic neural network with programmable phase shifters and optical power monitoring to solve classification tasks using in situ backpropagation. Experiments performed comparably to digital simulations with over 94% test accuracy and the energy scaling analysis indicated a route to scalable machine learning systems. The chip incorporates a photonic accelerator that allows calculations to be carried out very quickly and efficiently. The advantages of photonic neural networks have long been known, but the team has now succeeded in implementing training strategies for photonic neural networks.