Small-scale photonic Kolmogorov-Arnold networks using standard telecom nonlinear modules
2026-04-09 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors developed a new type of small photonic neural network that uses light-based components entirely, avoiding slower electronic parts. Their design uses special optical modules with controllable elements to create nonlinear behavior needed for learning tasks. Even with just a few modules, their network performs well on tasks like classification and image recognition, nearly matching software-based models but with fewer parts. They also showed their design works well even when the input signals are noisy or of low precision. Their approach uses realistic physics models to optimize these optical networks, making practical experiments with standard telecom hardware more achievable.
photonic neural networknonlinear opticsMach-Zehnder interferometersemiconductor optical amplifieroptical attenuatorKolmogorov-Arnold networkoptical signal processingend-to-end optimizationtelecommunications hardware
Authors
Luca Nogueira Calçado, Sergei K. Turitsyn, Egor Manuylovich
Abstract
Photonic neural networks promise ultrafast inference, yet most architectures rely on linear optical meshes with electronic nonlinearities, reintroducing optical-electrical-optical bottlenecks. Here we introduce small-scale photonic Kolmogorov-Arnold networks (SSP-KANs) implemented entirely with standard telecommunications components. Each network edge employs a trainable nonlinear module composed of a Mach-Zehnder interferometer, semiconductor optical amplifier, and variable optical attenuators, providing a four-parameter transfer function derived from gain saturation and interferometric mixing. Despite this constrained expressivity, SSP-KANs comprising only a few optical modules achieve strong nonlinear inference performance across classification, regression, and image recognition tasks, approaching software baselines with significantly fewer parameters. A four-module network achieves 98.4\% accuracy on nonlinear classification benchmarks inaccessible to linear models. Performance remains robust under realistic hardware impairments, maintaining high accuracy down to 6-bit input resolution and 14 dB signal-to-noise ratio. By using a fully differentiable physics model for end-to-end optimisation of optical parameters, this work establishes a practical pathway from simulation to experimental demonstration of photonic KANs using commodity telecom hardware.