Programmable phase-change metasurfaces on waveguides for multimode photonic convolutional neural network

• Published in: Nature communications Vol. 12; no. 1; p. 96
• Main Authors: Wu, Changming, Yu, Heshan, Lee, Seokhyeong, Peng, Ruoming, Takeuchi, Ichiro, Li, Mo
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 04.01.2021; Nature Publishing Group; Nature Portfolio
• Subjects: 132/122; 639/624/399/1099; 639/925/927/1021; Algorithms; Artificial neural networks; Bandwidths; Computation; Converters; Data transmission; Digital electronics; Humanities and Social Sciences; Image contrast; Image processing; Learning algorithms; Machine learning; Mathematical analysis; Matrix algebra; Matrix methods; Metasurfaces; Microprocessors; multidisciplinary; Multiplication; Neural networks; Object recognition; Phase change materials; Phase transitions; Photonics; Refractivity; Science; Science (multidisciplinary); Waveguides
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/s41467-020-20365-z

Neuromorphic photonics has recently emerged as a promising hardware accelerator, with significant potential speed and energy advantages over digital electronics for machine learning algorithms, such as neural networks of various types. Integrated photonic networks are particularly powerful in performing analog computing of matrix-vector multiplication (MVM) as they afford unparalleled speed and bandwidth density for data transmission. Incorporating nonvolatile phase-change materials in integrated photonic devices enables indispensable programming and in-memory computing capabilities for on-chip optical computing. Here, we demonstrate a multimode photonic computing core consisting of an array of programable mode converters based on on-waveguide metasurfaces made of phase-change materials. The programmable converters utilize the refractive index change of the phase-change material Ge 2 Sb 2 Te 5 during phase transition to control the waveguide spatial modes with a very high precision of up to 64 levels in modal contrast. This contrast is used to represent the matrix elements, with 6-bit resolution and both positive and negative values, to perform MVM computation in neural network algorithms. We demonstrate a prototypical optical convolutional neural network that can perform image processing and recognition tasks with high accuracy. With a broad operation bandwidth and a compact device footprint, the demonstrated multimode photonic core is promising toward large-scale photonic neural networks with ultrahigh computation throughputs. Integrated optical computing requires programmable photonic and nonlinear elements. The authors demonstrate a phase-change metasurface mode converter, which can be programmed to control the waveguide mode contrast, and build an optical convolutional neural network to perform image processing tasks.