An official website of the United States government
Optical neural networks (ONNs), implemented on an array of cascaded Mach–Zehnder interferometers (MZIs), have recently been proposed as a possible replacement for conventional deep learning hardware. They potentially offer higher energy efficiency and computational speed when compared to their electronic counterparts. By utilizing tunable phase shifters, one can adjust the output of each of MZI to enable emulation of arbitrary matrix–vector multiplication. These phase shifters are central to the programmability of ONNs, but they require a large footprint and are relatively slow. Here we propose an ONN architecture that utilizes parity–time (PT) symmetric couplers as its building blocks. Instead of modulating phase, gain–loss contrasts across the array are adjusted as a means to train the network. We demonstrate that PT symmetric ONNs (PT-ONNs) are adequately expressive by performing the digit-recognition task on the Modified National Institute of Standards and Technology dataset. Compared to conventional ONNs, the PT-ONN achieves a comparable accuracy (67% versus 71%) while circumventing the problems associated with changing phase. Our approach may lead to new and alternative avenues for fast training in chip-scale ONNs.
more »
« less
This content will become publicly available on July 6, 2026
A Physics-Aware Simulation Platform for Phase-Only Optical Neural Networks
A physics-aware simulation platform is proposed for optical neural networks (ONNs), incorporating phase-only modulation and passive free-space propagation. The platform enables end-to-end training under experimentally realistic constraints, with both the phase mask and propagation distance treated as learnable parameters. To facilitate classification, structured 2D output patterns are introduced, where each label corresponds to a fixed spatial light spot. When evaluated with the MNIST dataset, the system achieves 94.6% accuracy using a single phase modulation layer, demonstrating the effectiveness of spatial encoding in physically plausible ONNs.
more »
« less
- Award ID(s):
- 2211990
- PAR ID:
- 10631843
- Publisher / Repository:
- IEEE
- Date Published:
- ISBN:
- 979-8-3315-9777-1
- Page Range / eLocation ID:
- 1 to 4
- Format(s):
- Medium: X
- Location:
- Barcelona, Spain
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
A photonic Bragg grating is a fundamental building block that reflects the direction of wave propagation through spatial phase modulation and can be implemented using sidewall corrugation. However, due to the asymmetric aspect ratio of a waveguide cross section, typical Bragg gratings exhibit a strong polarization sensitivity. Here, we show that photonic Bragg gratings with cladding asymmetry can enable polarization-independent notch filters by rotating input polarizations. Such Bragg gratings strongly couple transverse electric (TE) and transverse magnetic (TM) modes propagating in opposite directions, filtering the input signal and reflecting the rotated mode. We analyzed this polarization-rotating Bragg grating using the coupled-mode theory and experimentally demonstrated it on a silicon-on-insulator platform. Our device concept is simple to implement and compatible with other platforms, readily available as polarization transparent Bragg components.more » « less
-
We present a general framework capable of describing the nonlinear propagation of pulsed optical beams of arbitrary shapes and phase fronts inside a graded-index (GRIN) fiber. The main assumption made is that the spatial self-imaging features of the beam are not affected by the temporal evolution of optical pulses. A propagation kernel known from the work done in the 1970s is used to obtain a distance-dependent nonlinear coefficient that captures all spatial effects within an effective nonlinear Schrödinger equation. We consider three specific beam shapes (Gaussian, circular, and square) to study the impact of the shape, position, and curvature of optical beams on the complex spatiotemporal dynamics specific to GRIN fibers. In particular, we focus on the impact of an input beam’s shape on the modulation-instability sidebands and the generation of multiple dispersive waves when higher-order solitons form inside a GRIN fiber. The results of our numerical analysis indicate that for beam widths chosen to yield the same value of the effective mode area at the input end of the fiber, the nonlinear effects are pronounced considerably when a Gaussian beam is launched into the fiber. We also found that even though the self-imaging period is doubled when an off-centered Gaussian beam is launched into a GRIN fiber, it does not affect the nonlinear evolution because the effective beam area still maintains the same periodicity, as long as the shift in the beam’s center is not so large that it does not remain confined to the fiber’s core.more » « less
-
Machine learning with artificial neural networks has recently transformed many scientific fields by introducing new data analysis and information processing techniques. Despite these advancements, efficient implementation of machine learning on conventional computers remains challenging due to speed and power constraints. Optical computing schemes have quickly emerged as the leading candidate for replacing their electronic counterparts as the backbone for artificial neural networks. Some early integrated photonic neural network (IPNN) techniques have already been fast-tracked to industrial technologies. This review article focuses on the next generation of optical neural networks (ONNs), which can perform machine learning algorithms directly in free space. We have aptly named this class of neural network model the free space optical neural network (FSONN). We systematically compare FSONNs, IPNNs, and the traditional machine learning models with regard to their fundamental principles, forward propagation model, and training process. We survey several broad classes of FSONNs and categorize them based on the technology used in their hidden layers. These technologies include 3D printed layers, dielectric and plasmonic metasurface layers, and spatial light modulators. Finally, we summarize the current state of FSONN research and provide a roadmap for its future development.more » « less
-
Abstract Recent advances in photonic integrated circuits have enabled a new generation of programmable Mach–Zehnder meshes (MZMs) realized by using cascaded Mach–Zehnder interferometers capable of universal linear-optical transformations onNinput/output optical modes. MZMs serve critical functions in photonic quantum information processing, quantum-enhanced sensor networks, machine learning and other applications. However, MZM implementations reported to date rely on thermo-optic phase shifters, which limit applications due to slow response times and high power consumption. Here we introduce a large-scale MZM platform made in a 200 mm complementary metal–oxide–semiconductor foundry, which uses aluminium nitride piezo-optomechanical actuators coupled to silicon nitride waveguides, enabling low-loss propagation with phase modulation at greater than 100 MHz in the visible–near-infrared wavelengths. Moreover, the vanishingly low hold-power consumption of the piezo-actuators enables these photonic integrated circuits to operate at cryogenic temperatures, paving the way for a fully integrated device architecture for a range of quantum applications.more » « less
