(2018)All-optical machine learning using diffractive deep neural networks “All-optical machine learning using diffractive deep neural networks”,这篇Science上的文章发表于2018年6月{Lin X, Rivenson Y, Yardimci N T, et al
而在衍射神经网络中,这两个数据(权重和偏置)与的两个部分紧密相关:光相位和幅值的改变以及衍射层的透射和反射系数。光相位和幅度的改变是由瑞利-索末菲衍射积分公式所确定的,而偏置项是由层的透射系数所确定的。衍射神经网络模型如下图,下面来详细介绍一下结构。 关于神经元之间连接的权重,通过瑞利-索末菲衍射积...
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 ...
Here we introduce an all-optical deep learning framework, where the neural network is physically formed by multiple layers of diffractive surfaces that work in collaboration to optically perform an arbitrary function that the network can statistically learn. We term this framework as Diffractive Dee...
We utilized deep learning-based optimization with stochastic gradient descent to optimize the thickness values of the diffractive features on the diffractive layers. This training was targeted at minimizing a custom-designed loss function defined by the mean squared error (MSE) between the diffractive ...
A diffractive deep neural network is an optical machine learning structure that is capable of combining deep learning with optical diffraction and light-matter interaction to design diffractive surfaces that collectively carry out optical computation at the speed of light. Prototype of a broadband diffrac...
(i.e., neurons) of diffractive surfaces are adjusted or trained to perform a desired input–output transformation task as the light diffracts through these layers. Trained with deep-learning-based error back-propagation methods, these diffractive networks have been shown to perform machine-learning ...
(see Section 2 in Supple- mentary Materials), the phase and amplitude of the cross- polarized transmitted light are customized independently with a high spatial precision of subwavelength scale, hereby superior to traditional diffractive optical elements43, spatial light modulators44, and digital ...
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...
In our second implementation (Fig.6), we performed four different, arbitrary linear transformations (i.e.,Np= 4) using a diffractive network composed of eight transmissive layers that are jointly optimized using deep learning and examples of input/output fields corresponding to the selected complex...