Physics‐Aware Machine Learning and Adversarial Attack in Complex‐Valued Reconfigurable Diffractive All‐Optical Neural Network
Physics‐Aware Machine Learning and Adversarial Attack in Complex‐Valued Reconfigurable Diffractive All‐Optical Neural Network
复制标题
物理 - 复杂的感知机器学习和对抗性攻击 - 有价值的可重构衍射全 - 光神经网络
DOI：
10.1002/lpor.202200348
复制
发表时间：
2022
影响因子：
11
通讯作者：
Gao, Weilu
中科院分区：
文献类型：
--
作者：
Chen, Ruiyang;Li, Yingjie;Lou, Minhan;Fan, Jichao;Tang, Yingheng;Sensale‐Rodriguez, Berardi;Yu, Cunxi;Gao, Weilu
衍射光学神经网络在加速现代机器学习（ML）算法方面比电子电路显示出有希望的优势。然而，实现完全可编程的全光实现和快速硬件部署是具有挑战性的。本文提出了一种基于级联传输扭曲向列液晶空间光调制器的大规模、高性价比、复杂价值和可重构的衍射全光神经网络系统。分类重参数化技术的使用创建了一个物理感知的训练框架，用于将计算机训练的模型快速准确地部署到光学硬件上。这样一个完整的硬件和软件堆栈不仅可以在标准数据集中进行手写体数字分类的实验演示，还可以对系统的物理感知对抗性攻击进行理论分析和实验验证，这些攻击是由基于复值梯度的算法生成的。通过与传统多层感知器和卷积神经网络的详细对抗性鲁棒性比较，衍射光学神经网络具有明显的统计对抗性。开发的全栈软件和硬件为在各种机器学习任务和光学对抗性机器学习研究中使用衍射光学提供了新的机会。
Diffractive optical neural networks have shown promising advantages over electronic circuits for accelerating modern machine learning (ML) algorithms. However, it is challenging to achieve fully programmable all‐optical implementation and rapid hardware deployment. Here, a large‐scale, cost‐effective, complex‐valued, and reconfigurable diffractive all‐optical neural networks system in the visible range is demonstrated based on cascaded transmissive twisted nematic liquid crystal spatial light modulators. The employment of categorical reparameterization technique creates a physics‐aware training framework for the fast and accurate deployment of computer‐trained models onto optical hardware. Such a full stack of hardware and software enables not only the experimental demonstration of classifying handwritten digits in standard datasets, but also theoretical analysis and experimental verification of physics‐aware adversarial attacks onto the system, which are generated from a complex‐valued gradient‐based algorithm. The detailed adversarial robustness comparison with conventional multiple layer perceptrons and convolutional neural networks features a distinct statistical adversarial property in diffractive optical neural networks. The developed full stack of software and hardware provides new opportunities of employing diffractive optics in a variety of ML tasks and in the research on optical adversarial ML.
登录
查看更多内容
影响因子：
4.6
作者：
Li Y;Chen R;Sensale-Rodriguez B;Gao W;Yu C
通讯作者：
Yu C
影响因子：
16.6
作者：
Wang T;Ma SY;Wright LG;Onodera T;Richard BC;McMahon PL
通讯作者：
McMahon PL
影响因子：
35
作者：
Shen, Yichen;Harris, Nicholas C.;Soljacic, Marin
通讯作者：
Soljacic, Marin
影响因子：
19.4
作者：
Overvig, Adam C.;Shrestha, Sajan;Yu, Nanfang
通讯作者：
Yu, Nanfang
影响因子：
35
作者：
S. Rodrigues;Ziqi Yu;P. Schmalenberg;J. Lee;H. Iizuka;E. Dede
通讯作者：
E. Dede