In article number 2000249, Jian Wang and co‐workers show the intelligent recognition of various spatial mode bases by convolutional neural network (CNN)‐based deep learning. Four mode bases in a ring‐core fiber, i.e. linearly polarized mode group, linearly polarized orbital angular momentum (OAM) mode group, circularly polarized OAM mode group, and vector mode group, are successfully recognized with high recognition rate. This may open a door for deep learning enabled robust and intelligent optical communications exploiting diverse spatial modes for sustainable capacity scaling.

中文翻译：

---

深度学习：环芯光纤中基于深度学习的不同模式基础的识别（Laser Photonics Rev. 14（11）/ 2020）

Wang Jian及其同事在文章2000249中展示了基于卷积神经网络（CNN）的深度学习对各种空间模式基础的智能识别。环形芯光纤中的四个模式基，即线性偏振模式组，线性偏振轨道角动量（OAM）模式组，圆偏振OAM模式组和矢量模式组，已被成功识别，识别率很高。这可能为利用深度学习实现的强大而智能的光通信打开一扇门，该通信利用各种空间模式来实现可持续的容量缩放。