Statistical Computing with Integrated Photonics System

Artificial Neural Networks (ANN) have dramatically improved performance for many machine learning tasks. Recently, special hardware has been developed to carry out this particular type of algorithms. We demonstrate new architectures based on Photonic Integrated Circuits for carrying out ANN and other statistical processing tasks. We will discuss the challenges and opportunities of such approach, and how it can enable significant enhancement in computational speed, latency and power efficiency over state of the arts electronics.