Abstract:Optical convolution computing is gaining traction owing to its inherent parallelism, multi-dimensional processing, and energy efficiency. To handle input dimensions of N, conventional implementations necessitate N2 optical elements, such as Mach-Zehnder interferometers or micro-ring resonators, to process multiply-accumulate (MAC) operations, limiting scalability and resulting in elevated power consumption. Here, a direct convolution computing method based on wavelength routing, utilizing the unique sliding property of an arrayed waveguide grating router (AWGR) to perform the sliding window operation of the convolution in the wavelength-space domains is proposed. With two input vectors directly loaded onto two modulator arrays, the convolution result is instantaneously produced at a photodetector array. The entire convolution computation is executed within a single clock cycle without the need for preprocessing or decomposition into elementary MAC operations. The number of active elements is minimal, only needed for input/output. The proposed optical convolution unit has striking advantages of high scalability, high speed, and processing simplicity compared to those based on optical matrix-vector multipliers. In the first experimental demonstration, a remarkable classification accuracy of up to 98.2% in handwritten digit recognition tasks using a LeNet-5 neural network is achieved. A direct convolution computing method based on wavelength routing, utilizing the unique sliding property of an arrayed waveguide grating router (AWGR) to perform the sliding window operation of the convolution in the wavelength-space domains is proposed and demonstrated. With two inputs vectors directly loaded onto modulator arrays, the convolution result is instantaneously produced at a photodetector array. image

Optical Color Routing Enabled by Deep Learning

Full-color nanorouter for high-resolution imaging

Deep learning-optical network routing algorithm based on wavelength continuity supervision

Colour Routing with Single Silver Nanorods

Design of multifunctional color routers with Kerker switching using generative adversarial networks

Spatially Variant Deblur and Image Enhancement in a Single Multimode Fiber Imaged by Deep Learning

Inverse Design of Ultranarrow and High-Efficiency Color Filters Based on Tandem Convolutional Neural Networks

Direct Optical Convolution Computing Based on Arrayed Waveguide Grating Router

Deep Learning for Photonic Design and Analysis: Principles and Applications

Deep learning enabled inverse design in nanophotonics

Deep Neural Network-Based Inverse Design of a Novel Reconfigurable Multi-Tap Optical Filter

Mapping the design space of photonic topological states via deep learning

Subwavelength Color Router

Multi-task topology optimization of photonic devices in low-dimensional Fourier domain via deep learning

Subwavelength Bayer RGB Color Routers with Perfect Optical Efficiency

Deep learning: a new tool for photonic nanostructure design

A Routing Optimization Method for Software-Defined Optical Transport Networks Based on Ensembles and Reinforcement Learning

Deep Learning with Passive Optical Nonlinear Mapping

Nanophotonic structure inverse design for switching application using deep learning

Deep learning in nano-photonics: inverse design and beyond

Directional color routing assisted by switchable Fano resonance in bimetallic metagrating