Neural network learning with photonics and for photonic circuit design
Brunner, D.; Soriano, M. C.; Fan, S.
Nanophotonics 12, 773-775 (2023)
This special issue covers works that lie at the interface between machine learning, spearheaded by the computing power of artificial neural networks (NN), and photonic technologies. In the past few years, there has been a renewed interest in this promising field due to a number of successful experimental demonstrations of advanced computing functionalities or the design of optimized nanophotonic devices. Current trends in the community can be conceptually divided in two distinct research directions. On the one hand, photonic systems and devices can serve as a hardware substrate that naturally suits the characteristic properties of artificial NN topologies. Advantages brought by photonics in this context include the potential for parallelization, high-speed operation, and low power consumption. On the other hand, machine learning can aid in the design of photonic devices or components and accelerate the search for promising structures. Artificial NN can also assist in the processing of optically acquired data with the ultimate goal of adding new functionalities and enhancing performance.