arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2209.14252cs.AI
收藏

衍射光学神经网络的物理感知可微离散协同设计

Physics-aware Differentiable Discrete Codesign for Diffractive Optical Neural Networks

University of Utah(犹他大学)

Yingjie Li, Ruiyang Chen, Weilu Gao, Cunxi Yu

摘要:

与传统的深度神经网络(DNN)相比,衍射光学神经网络(DONN)在功率效率、并行性和计算速度方面具有显著的优势,因此引起了人们的广泛关注。然而,将算法训练的物理模型参数逆映射到具有离散值的真实世界光学设备上是一项重要的任务,因为现有的光学设备具有非统一的离散水平和非单调特性。这项工作提出了一种新的设备到系统的硬件-软件协同设计框架,它使有效的物理感知训练的DONN w.r.t任意实验测量的光学设备跨层。具体而言,Gumbel-Softmax用于实现从真实世界设备参数到DONN的前向函数的可微分离散映射,其中DONN中的物理参数可以通过简单地最小化ML任务的损失函数来训练。结果表明,我们提出的框架提供了显着的优势,传统的基于量化的方法,特别是与低精度的光学设备。最后,在低精度的物理实验光学系统的设置所提出的算法进行了充分验证。

英文摘要:

Diffractive optical neural networks (DONNs) have attracted lots of attention as they bring significant advantages in terms of power efficiency, parallelism, and computational speed compared with conventional deep neural networks (DNNs), which have intrinsic limitations when implemented on digital platforms. However, inversely mapping algorithm-trained physical model parameters onto real-world optical devices with discrete values is a non-trivial task as existing optical devices have non-unified discrete levels and non-monotonic properties. This work proposes a novel device-to-system hardware-software codesign framework, which enables efficient physics-aware training of DONNs w.r.t arbitrary experimental measured optical devices across layers. Specifically, Gumbel-Softmax is employed to enable differentiable discrete mapping from real-world device parameters into the forward function of DONNs, where the physical parameters in DONNs can be trained by simply minimizing the loss function of the ML task. The results have demonstrated that our proposed framework offers significant advantages over conventional quantization-based methods, especially with low-precision optical devices. Finally, the proposed algorithm is fully verified with physical experimental optical systems in low-precision settings.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑