A review of photonic integrated neural networks

Neural networks (NNs), inspired by the biological visual cortex, have been widely applied in computer vision, natural language processing, and beyond. As Moore's Law approaches its physical limits, photonic neural networks (PNNs), which leverage photons rather than electrons as information carriers, have emerged as a promising technological advancement. With ultrafast speeds and ultralow energy consumption in ultra-high throughput, PNNs usher in a new generation of intelligent computing. Moreover, the advent of photonic integrated circuits (PICs) provides a compact and reliable hardware platform for computing, further advancing the development of PNNs. In this paper, we review recent advances in linear and nonlinear computation and integrated optical devices serving as fan-in and fan-out. We also summarize large-scale on-chip PNN and discuss the future challenges associated with their monolithic integration.