Noise-resilient and high-speed deep learning with coherent silicon photonics.
Noise-resilient and high-speed deep learning with coherent silicon photonics.
复制标题
DOI：
10.1038/s41467-022-33259-z
复制
发表时间：
2022-09-23
影响因子：
16.6
通讯作者：
中科院分区：
文献类型：
--
作者：
深度学习应用的爆炸式增长引发了计算硬件的新时代，目标是有效部署乘法和累加运算。在这个领域，集成光子学已经成为一种有前途的节能深度学习技术平台，可以实现超高的计算速率。然而，尽管集成光子神经网络布局已经成功渗透到深度学习时代，但其计算速率和噪声相关特性仍然远远超出了高速光子引擎的承诺。在本文中，我们通过实验展示了一种噪声弹性深度学习相干光子神经网络布局，该布局以10 GMAC/sec/轴突的计算速率运行，并遵循噪声弹性训练模型。相干光子神经网络已被制造为硅光子芯片，其MNIST分类性能进行了实验评估，分别在5和10 GMAC/sec/axon下支持>99%和>98%的准确度值，提供了6倍的片上计算速率和>7%的准确度改进。深度学习应用中高速、高精度相干光子神经元的挑战在于解决噪声相关问题。在这里，Albergias-Alexandris等人通过引入抗噪声硬件架构和深度学习训练平台来解决这个问题。
The explosive growth of deep learning applications has triggered a new era in computing hardware, targeting the efficient deployment of multiply-and-accumulate operations. In this realm, integrated photonics have come to the foreground as a promising energy efficient deep learning technology platform for enabling ultra-high compute rates. However, despite integrated photonic neural network layouts have already penetrated successfully the deep learning era, their compute rate and noise-related characteristics are still far beyond their promise for high-speed photonic engines. Herein, we demonstrate experimentally a noise-resilient deep learning coherent photonic neural network layout that operates at 10GMAC/sec/axon compute rates and follows a noise-resilient training model. The coherent photonic neural network has been fabricated as a silicon photonic chip and its MNIST classification performance was experimentally evaluated to support accuracy values of >99% and >98% at 5 and 10GMAC/sec/axon, respectively, offering 6× higher on-chip compute rates and >7% accuracy improvement over state-of-the-art coherent implementations. The challenge of high-speed and high-accuracy coherent photonic neurons for deep learning applications lies to solve noise related issues. Here, Mourgias-Alexandris et al. address this problem by introducing a noise-resilient hardware architectural and a deep learning training platform.
登录
查看更多内容
影响因子：
8.6
作者：
RECK, M;ZEILINGER, A;BERTANI, P
通讯作者：
BERTANI, P
影响因子：
2.9
作者：
Semenova, N.;Porte, X.;Brunner, D.
通讯作者：
Brunner, D.
影响因子：
3.2
作者：
Stabile, R.;Dabos, G.;Pleros, N.
通讯作者：
Pleros, N.
DOI：
10.1109/tetci.2019.2923001
发表时间：
2021-06-01
影响因子：
5.3
作者：
Passalis, Nikolaos;Mourgias-Alexandris, George;Tefas, Anastasios
通讯作者：
Tefas, Anastasios
影响因子：
4.7
作者：
Tait, Alexander N.;Nahmias, Mitchell A.;Prucnal, Paul R.
通讯作者：
Prucnal, Paul R.