Wang Zi, Chang Lorry, Wang Feifan, Li Tiantian, Gu Tingyi
Department of Electrical and Computer Engineering, University of Delaware, Newark, DE, 19711, USA.
Nat Commun. 2022 Apr 19;13(1):2131. doi: 10.1038/s41467-022-29856-7.
Miniaturized image classifiers are potential for revolutionizing their applications in optical communication, autonomous vehicles, and healthcare. With subwavelength structure enabled directional diffraction and dispersion engineering, the light propagation through multi-layer metasurfaces achieves wavelength-selective image recognitions on a silicon photonic platform at telecommunication wavelength. The metasystems implement high-throughput vector-by-matrix multiplications, enabled by near 10 nanoscale phase shifters as weight elements within 0.135 mm footprints. The diffraction manifested computing capability incorporates the fabrication and measurement related phase fluctuations, and thus the pre-trained metasystem can handle uncertainties in inputs without post-tuning. Here we demonstrate three functional metasystems: a 15-pixel spatial pattern classifier that reaches near 90% accuracy with femtosecond inputs, a multi-channel wavelength demultiplexer, and a hyperspectral image classifier. The diffractive metasystem provides an alternative machine learning architecture for photonic integrated circuits, with densely integrated phase shifters, spatially multiplexed throughput, and data processing capabilities.
小型化图像分类器有潜力彻底改变其在光通信、自动驾驶车辆和医疗保健领域的应用。通过亚波长结构实现定向衍射和色散工程，光在多层超表面中的传播在电信波长的硅光子平台上实现了波长选择性图像识别。这些超系统通过在0.135平方毫米的面积内使用近10个纳米级移相器作为权重元件，实现了高通量的向量与矩阵乘法。衍射所体现的计算能力包含了与制造和测量相关的相位波动，因此经过预训练的超系统无需后期调整就能处理输入中的不确定性。在此，我们展示了三种功能超系统：一种15像素的空间模式分类器，对飞秒输入的准确率接近90%，一种多通道波长解复用器，以及一种高光谱图像分类器。衍射超系统为光子集成电路提供了一种替代的机器学习架构，具有密集集成的移相器、空间复用的吞吐量和数据处理能力。
