With recent rapid advances in photonic integrated circuits, it has been demonstrated that programmable photonic chips can be used to implement artificial neural networks. Convolutional neural networks (CNN) are a class of deep learning methods that have been highly successful in applications such as image classification and speech processing. We present an architecture to implement a photonic CNN using the Fourier transform property of integrated star couplers. We show, in computer simulation, high accuracy image classification using the MNIST dataset. We also model component imperfections in photonic CNN and show that the performance degradation can be recovered in a programmable chip. Our proposed architecture provides a large reduction in physical footprint compared to current implementations as it utilizes the natural advantages of optics and hence offers a scalable pathway towards integrated photonic deep learning processors.
Artificial intelligence (AI) techniques have been spreading in most scientific areas and have become a heated focus in photonics research in recent years. Forward modeling and inverse design using AI can achieve high efficiency and accuracy for photonics components. With AI-assisted electronic circuit design for photonics components, more advanced photonics applications have emerged. Photonics benefit a great deal from AI, and AI, in turn, benefits from photonics by carrying out AI algorithms, such as complicated deep neural networks using photonics components that use photons rather than electrons. Beyond the photonics domain, other related research areas or topics governed by Maxwell’s equations share remarkable similarities in using the help of AI. The studies in computational electromagnetics, the design of microwave devices, as well as their various applications greatly benefit from AI. This article reviews leveraging AI in photonics modeling, simulation, and inverse design; leveraging photonics computing for implementing AI algorithms; and leveraging AI beyond photonics topics, such as microwaves and quantum-related topics.
Integrated photodetectors are key building blocks of scalable photonics platforms. Many recent improvements have been made for integrated avalanche photodetectors (APDs) operating at infrared telecommunications wavelengths, but their visible-spectrum counterparts remain relatively unexplored. Here, we demonstrate PIN-doped silicon APDs for visible light detection, monolithically integrated with a silicon nitride photonics circuit via end-fire coupling. An in-depth study of multiple PIN doping profiles reveals different optimal designs based on the desired operating regimes. At −49 dBm input power, they show 0.25 A/W (0.8 A/W) responsivity at reverse bias as low as 0.5 V (5.5 V), with corresponding dark current of $< $ 3 pA (50 pA). We also report fast RF response with an optimal 3 dB bandwidth of 11 GHz and gain-bandwidth product of 142 GHz, with all devices yielding open eye diagrams at 25 Gbps or above. Coupled with CMOS-compatible fabrication, our APDs will enable scalable photonics applications in sensing, communications, and quantum technologies at visible wavelengths.
A cascaded Mach-Zehnder interferometer based filter for coarse wavelength (de)multiplexing (CWDM) at the O-band is fabricated and tested on a silicon nitride on SOI platform. We characterize the chip-to-chip performance variability of the filter devices on a wafer. Using the optical measurement data, we apply Bayesian inference methods to estimate the waveguide geometery parameters and also quantify the uncertainty of the estimates.
23 November 2010 in Journal of the Optical Society of America B
Optics InfoBase is the Optical Society's online library for flagship journals, partnered and copublished journals, and recent proceedings from OSA conferences.