在这里,我们介绍了一种物理机制,通过演示全光学衍射深度神经网络(D2NN, diffractive deep neural network)架构来执行机器学习,该架构可以使用基于深度学习的设计的集成工作的被动衍射层实现各种功能。我们创建了3D打印的D2NN,它实现了手写数字和时尚产品图像的分类,以及太赫兹光谱成像镜头的功能。 我们的全光学深度学习框...
之前提到的衍射神经网络的运算都是在实空间完成的运算,而这篇论文展示了将图像进行傅里叶变换后再经过神经网络处理的方法。这种方法能够完成更加复杂的计算机视觉的任务(例如分割),同时也被证实了相比以往,傅里叶衍射神经网络在一些任务上有着更高的正确率和稳定性。 这篇论文中,作者使用傅里叶衍射神经网络,实现了计...
the Diffractive Deep Neural Network has been completed. The last part is training it, and I don't want to write down more details, which are intuitive, for convenience. Actually, Just using 2 layers, the
Figure1shows the proposed real-time multi-task diffractive deep neural network (D2NN) architecture. Specifically, in this work, our multi-task D2NN deploy image classification DNN algorithms with two tasks, i.e., classifying MNIST10 dataset and classifying Fashion-MNIST10 dataset. In a single-...
The development of artificial intelligence is typically focused on computer algorithms and integrated circuits. Recently, all-optical diffractive deep neural networks have been created that are based on passive structures and can perform complicated func
Diffractive Deep Neural Network (D${}^2$NN) Get Started OpticalLayer.py includes two types of Layers. Diffraction is the free space propagtion layer and DiffLayer provides phase modulation. from torch import nn from OpticalLayers import DiffLayer, Diffraction Build your own model like you do ...
Using a 3D printer, a research team at the UCLA Samueli School of Engineering has created an artificial neural network that can analyze large volumes of data and identify objects at the speed of light. Called a diffractive deep neural network (D2NN), the technology uses the light scattering ...
(2018)All-optical machine learning using diffractive deep neural networks,程序员大本营,技术文章内容聚合第一站。
Lin, “Optical multi-task learning using multi-wavelength diffractive deep neural networks,” Nanophotonics, vol. 12, no. 5, pp. 893–903, 2023. https://doi.org/10.1515/nanoph-2022-0615.Search in Google Scholar [7] T. Yan, et al.., “Fourier-space diffractive deep neural network,” ...
14. A method of diffractive deep neural network (D2NN) processing, the method comprising: receiving a light beam; performing, using a first modulation region of a modulation layer, a first spatial modulation of the light beam to obtain a first modulated signal; reflecting the first modulated ...