Photonic processor could enable ultrafast AI computations

Scientists from MIT and other institutions have developed a new photonic chip designed to overcome the limitations of traditional hardware in deep neural network processing. This fully integrated photonic processor performs all essential computations for a deep neural network optically on the chip. In a demonstration, the optical device completed the computations for a machine-learning classification task in less than half a nanosecond, achieving over 92% accuracy—comparable to conventional hardware performance.