智能光子处理系统：概念与研究进展

 

 

摘要：提出了智能光子处理系统（IPS）的概念，该系统包括人工智能（AI）赋能的IPS、光子辅助AI的IPS和神经拟态的IPS。AI赋能的IPS可实现更优的光子/光学系统信号处理性能；光子辅助AI的IPS重点研究高速率、低能耗的AI计算加速器；神经拟态的IPS将模拟人脑信息的处理机制，同时具备适应和决策能力。回顾了IPS的最新研究进展，并展望了其未来发展的挑战和机遇。
关键词：IPS；深度学习；人工智能；光学神经网络；神经拟态光子学


Abstract: In this paper, an intelligent photonic processing system (IPS) is proposed, including artificial intelligence (AI)-powered IPS, IPS with photonics-facilitated AI, and neuromorphic IPS. AI-powered IPS enables better signal processing performance in photonic/optical systems. The IPS with photonics-facilitated AI focuses on the computing accelerator with high speed and low power consumption. Inspired from the brain, the neuromorphic IPS features adaptation and decision-making abilities. The latest research of IPS are reviewed and some prospects for the future development are provided.
Keywords: IPS; deep learning; artificial intelligence; optical neural network; neuromorphic photonics