Optical Speckle Patterns as Nonlinear Feature Maps for Photonic Machine Learning

Kiia Arola; Mikko J. Huttunen

Home/Archive/arXiv

arXiv · 2610.35546

Optical Speckle Patterns as Nonlinear Feature Maps for Photonic Machine Learning

Kiia Arola·Mikko J. Huttunen

View original source ↗Read source document ↗Export metadata ↓

Abstract

The rapid growth of artificial intelligence (AI) has increased the demand for energy-efficient hardware, and optical computing has emerged as a promising alternative to conventional electronic processors owing to its high parallelism and potentially lower power consumption. In this work, we investigate an optical computing architecture based on an integrated photonic circular multipass cell, focusing on its suitability for machine learning and pattern recognition tasks. The properties of the system are evaluated through measures of nonlinearity and cosine similarity, which characterize the system's information-processing capacity and the dimensionality of the generated feature representations. Numerical results demonstrate that the reservoir mirror reflectance, the size of the encoded figure, and the refractive-index contrast between the reservoir and the encoded figure modify the nonlinearity of the system, as verified via pairwise similarity estimates. These results indicate that the features generated by the circular multipass cell can be tailored by adjusting the system parameters, providing a means to control the reservoir transformation of the input data. The resulting representations were applied to classification tasks involving texture images and thyroid collagen images, achieving classification accuracies ranging from 69% to 93%. These findings suggest that photonic computing systems could provide a basis for more energy-efficient machine-learning hardware.

Explore related subjects

physics.optics

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Kiia Arola, Mikko J. Huttunen. 2026-10-26. Optical Speckle Patterns as Nonlinear Feature Maps for Photonic Machine Learning. https://arxiv.org/abs/2610.35546

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Interaction-Induced Topological Transport of Extended Doublons

Topological boundary states are a defining feature of topological systems, providing robust transport protected against perturbations. While such states are commonly engineered in non-interacting systems, how to realize interaction-induced topological states composed of only few particles has remained experimentally unexplored thus far. In this work, we realize such states using a driven photonic lattice as a linear analogue platform. We observe the formation of spatially extended bound states of two interacting particles, in which the particles remain correlated while occupying different lattice sites. At discrete values of the local interaction, these bound states support robust transport governed by a topological phase arising from drive-induced correlated hopping. Our results establish an experimentally accessible route to interaction-driven topology beyond single-particle physics and provide a versatile framework for exploring correlated few-particle phenomena across different physical platforms.

physics.optics↗

arXiv2026

Recovering topological information of light by topological learning

The evolution of modern-day communication networks towards enhanced capacity and robustness is driving interest in topological light wave as a next-generation information carrier. However, harnessing its potential is hindered by the lack of an effective topological reader, particularly when strong disorder renders conventional readout intractable. Here we address this by introducing TOPO$^{2}$, an artificial intelligence framework that leverages on topological learning to recover topological information from light. We demonstrate it on highly distorted optical fields with embedded Skyrmion topologies, showing that we can extract the topological numbers in a single shot with just a conventional camera. We showcase its capability in a TOPO$^{2}$-based communication scheme, sending and recovering information encoded in a topological alphabet, confirming that it outperforms conventional approaches. Our connection of topological photonics with topological AI as a powerful paradigm for robust high-capacity information encoding and processing opens a new path towards topological information optics.

physics.optics↗

arXiv2026

Video-rate quantitative phase and polarimetric imaging (QP2I) for label-free imaging of contracting muscle cells

Quantitative phase imaging (QPI) has become a standard approach for label-free microscopy of transparent specimens, mapping the optical path length. However, the isotropic phase is blind to the molecular order underlying optical anisotropy. Implementations that also add quantitative contrast to birefringence rely on multi-frame illumination or polarization-state scanning, or on dedicated interferometric hardware and thus trade away acquisition speed or the simplicity of a standard microscope. We present Quantitative Phase and Polarimetric Imaging (QPPI), a computational imaging framework that converts a standard differential interference contrast (DIC) microscope with a polarization camera into a single-shot, video-rate, dual-contrast imaging system. We developed a new globally regularized phase reconstruction that eliminates artifacts of prior DIC-to-QPI approaches, providing quantitative phase information, and used per-pixel Stokes-based reconstruction to obtain quantitative anisotropic retardance. We validate the approach across simulations, static reference targets, fixed muscle tissue and living dynamically moving cells. QPPI enables all-optical and label-free studies of contractile dynamics in live cardiomyocytes. It provides both structural and functional readouts at sufficient frame rates to resolve fast cardiomyocyte contractions. Conventional fluorescent labels bleach over time, perturb physiology, and preclude long-term monitoring. Instead, QPPI enables long-term, label-free interrogation of dynamic cellular processes at quantitative contrast and provides structural readouts similar to those provided by established fluorescent labels, such as frame-by-frame sarcomere length. It also offers previously inaccessible measurements, like quantitative analysis of retardance changes during muscle contraction, which provides a direct optical signature of muscle photoelasticity.

physics.optics↗