[TuG3] Algorithms for Photonic Computing

3:45 PM - 4:00 PM JST(6:45 AM - 7:00 AM UTC)

[TuG3-7] Optical Diffraction Neural Network In-situ Clustering for Full Self-learning Ability

○Qiaomu Hu^1,2, Rui Zeng^1,2, Tiange Wu^1,2, Nuoyan Li^1,2, Yunlong Li^1,2, Yiqiang Ou^1,2, Shuang Zheng^1,2,3, Minming Zhang^1,2,3 (1. Huazhong University of Science & Technology (China), 2. National Engineering Research Center for Next Generation Internet Access System (China), 3. Optics Valley Laboratory (China))

Keywords:

Photonics in neuromorphic computing and machine learning devices,Highly parallelized scalable photonic computing architectures and devices

We proposed an in-situ unsupervised learning method for training optical diffraction neural networks. Once deployed, our network achieved an image recognition accuracy of 88.37% without image labels or human intervention.