Multi-channel Optical Vision Model
Abstract
Spatial multiplexing is one of the natural strengths of optics, yet in optical neural networks, it is often used mainly as parallel throughput. Here, we show that spatial multiplexing in an optical neural network can be used not only to process multiple inputs in parallel, but also to define a trainable representational coordinate of the model. In three implemented scenarios, parallel-input processing, class-code readout and channel-mixed feature interaction, spatial channels act as independent learners, structured code dimensions, and interacting feature groups. The programmable free-space optical processor is trained through an online physical-forward/surrogate-backward scheme, where measured optical outputs define the forward pass while a differentiable surrogate estimates gradients and is continually fine-tuned during training from newly acquired optical data. We demonstrate these channel roles in image classification and regression tasks using multi-layer architectures with more than one million trainable optical phase parameters. We further implement a hybrid optical-electronic vision-language model, in which the optical neural network provides visual tokens to a digital transformer decoder for controlled image-captioning tasks. These results establish spatially multiplexed optical channels as a programmable feature and readout space for hybrid optical vision models.
Related
- Continuous vision transformer-based neural operator for scalable beam propagation modeling in optical fibers
- Key Evaluation of a Relativistic Quantum Key Distribution Over a Turbulent Free-space Optical Channel
- Analytical Channel Modeling and Stability Aware Optimization of Optical Inter Satellite Links
- Performance Analysis of Multiparty GHZ-State Quantum Conference Key Agreement over Optical Fiber Channel
- Multi-Channel Non-Abelian Edge Pumping via Multi-Degeneracy Encircling
BibTeX
@misc{momeni2026multichannel,
title = {{Multi-channel Optical Vision Model}},
author = {Ali Momeni and Guillaume Noetinger and Tim Tuuva and Romain Fleury},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2606.10253},
eprint = {2606.10253},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2606.10253},
}
RIS
TY - UNPB
TI - Multi-channel Optical Vision Model
AU - Ali Momeni
AU - Guillaume Noetinger
AU - Tim Tuuva
AU - Romain Fleury
PY - 2026
DA - 2026/09/07
PB - arXiv
DO - 10.48550/arxiv.2606.10253
UR - https://arxiv.org/abs/2606.10253
AB - Spatial multiplexing is one of the natural strengths of optics, yet in optical neural networks, it is often used mainly as parallel throughput. Here, we show that spatial multiplexing in an optical neural network can be used not only to process multiple inputs in parallel, but also to define a trainable representational coordinate of the model. In three implemented scenarios, parallel-input processing, class-code readout and channel-mixed feature interaction, spatial channels act as independent learners, structured code dimensions, and interacting feature groups. The programmable free-space optical processor is trained through an online physical-forward/surrogate-backward scheme, where measured optical outputs define the forward pass while a differentiable surrogate estimates gradients and is continually fine-tuned during training from newly acquired optical data. We demonstrate these channel roles in image classification and regression tasks using multi-layer architectures with more than one million trainable optical phase parameters. We further implement a hybrid optical-electronic vision-language model, in which the optical neural network provides visual tokens to a digital transformer decoder for controlled image-captioning tasks. These results establish spatially multiplexed optical channels as a programmable feature and readout space for hybrid optical vision models.
ER -