An official website of the United States government
This content will become publicly available on March 30, 2026
Scalable Machine Learning Models for Optical Transmission System Management
Optical transmission systems require accurate modeling and performance estimation for autonomous adaption and reconfiguration. We present efficient and scalable machine learning (ML) methods for modeling optical networks at component- and network-level with minimized data collection.
more »
« less
- PAR ID:
- 10640956
- Publisher / Repository:
- Optica Publishing Group
- Date Published:
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
A prime objective of modeling optical fibers is capturing mode confinement losses correctly. This paper demonstrates that specific modeling choices, especially regarding the outer fiber cladding regions and the placement of the computational boundary, have significant impacts on the calculated mode losses. This sensitivity of the computed mode losses is especially high for microstructure fibers that do not guide light by total internal reflection. Our results illustrate that one can obtain disparate mode confinement loss profiles for the same optical fiber design simply by moving the boundary to a new material region. We conclude with new recommendations for how to better model these losses.more » « less
-
We propose a workflow for modeling generalized mid-spatial frequency (MSF) errors in optical imaging systems. This workflow enables the classification of MSF distributions, filtering of bandlimited signatures, propagation of MSF errors to the exit pupil, and performance predictions that differentiate performance impacts due to the MSF distributions. We demonstrate the workflow by modeling the performance impacts of MSF errors for both transmissive and reflective imaging systems with near-diffraction-limited performance.more » « less
-
All-optical neural networks (AONNs) promise transformative gains in speed and energy efficiency for artificial intelligence (AI) by leveraging light's intrinsic parallelism and wave nature. However, their scalability has been fundamentally limited by the high power requirements of conventional nonlinear optical elements. Here, we present a low-power nonlinear activation scheme based on a three-level quantum system driven by dual laser fields. This platform introduces a two-channel nonlinear activation matrix with self- and cross-nonlinearities, enabling true multi-input, multi-output optical processing. The system supports tunable activation behaviors, including sigmoid and ReLU functions, at ultralow power levels (17μW per neuron). We validate our approach through theoretical modeling and experimental demonstration in rubidium vapor cells, showing the feasibility of scaling to deep AONNs with millions of neurons operating under 20 W of total optical power. Crucially, we demonstrate the all-optical generation of gradient-like signals with backpropagation, paving the way for all-optical training. These results mark a significant advance toward scalable, high-speed, and energy-efficient optical AI hardware.more » « less
