EUVIMEDEUVIMED entdecken
Open Access

Few-shot neuromorphic vision in a nonlinear photonic network laser

Wai Kit Ng iD, Jakub Dranczewski iD, Anna Fischer iD, T V Raziman iD, Dhruv Saxena iD, Tobias Farchy iD, Kilian Stenning iD, Jonathan Peters iD, Heinz Schmid iD, Will R Branford iD, Mauricio Barahona iD, Kirsten Moselund iD und 2 weitere

Science Advances · 2026

DOI: 10.1126/sciadv.aec6546 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science Advances, 2026. Verfasst von Wai Kit Ng und 13 weiteren Autor:innen. Beteiligte Einrichtung: Blackett Laboratory, Department of Physics, Imperial College London, London, United Kingdom. Hauptthemen laut MEDLINE: Lasers, Neural Networks, Computer, Artificial Intelligence. Die Arbeit nennt 65 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: Engineering Research Centers, Eric and Wendy Schmidt, Engineering and Physical Sciences Research Council.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

With the growing prevalence of artificial intelligence (AI), demand increases for hardware that mimics the brain’s ability to extract structure from limited data. In the retina, ganglion cells detect features from sparse inputs via lateral inhibition, where neurons antagonistically suppress activity of neighboring cells. Biological neurons exhibit diverse heterogeneous nonlinear responses, linked to robust learning and strong performance in low-data regimes. Here, we introduce a retinally inspired photonic computing system where spatially competing lasing modes in a random network laser act as heterogeneous, inhibitively coupled neurons, enabling feature detection, few-shot classification, and segmentation. This silicon-compatible scheme harnesses heterogeneous excitatory and inhibitory nonlinear physical dynamics, which give rise to emergent photonic computing behavior, including parallel feature detection and strong performance when training data are scarce. We report 98.05 and 87.85% accuracy on MNIST and Fashion-MNIST, and 90.12% on BreakHis cancer diagnosis, outperforming software convolutional neural networks including EfficientNetV2 and the vision transformer ViT in few-shot and class-imbalanced regimes with training sets of up to several hundred images. We demonstrate combined segmentation and classification on the HAM10k skin lesion dataset, achieving DICE and Jaccard scores of 84.49 and 74.80%. These results demonstrate the potential of random lasing networks as nonlinear photonic learning systems and highlight the ability of heterogeneous nonlinear dynamics to support strong learning in challenging low-data scenarios.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

LasersNeural Networks, ComputerArtificial IntelligenceNonlinear DynamicsConvolutional Neural NetworksPhotonsHumansSoft ComputingNeuronsAlgorithms

Autor:innen und Institutionen

  • Blackett Laboratory, Department of Physics, Imperial College London, London, United Kingdom.
  • IBM Research Europe-Zürich, Säumerstrasse 4, Rüschlikon, 8803, Switzerland.
  • Department of Mathematics, Imperial College London, London, United Kingdom.
  • London Centre for Nanotechnology, Imperial College London, London, United Kingdom.
  • Laboratory of Nano and Quantum Technologies, Paul Scherrer Institut, Switzerland.
  • INPhO, Faculty of Engineering, Ecole Polytechnique Fédérale de Lausanne, Switzerland.

ORCID-Profile: Wai Kit Ng, Jakub Dranczewski, Anna Fischer, T V Raziman, Dhruv Saxena, Tobias Farchy, Kilian Stenning, Jonathan Peters, Heinz Schmid, Will R Branford, Mauricio Barahona, Kirsten Moselund, Riccardo Sapienza, Jack C Gartside

Förderung

  • Engineering Research Centers
  • Eric and Wendy Schmidt
  • Engineering and Physical Sciences Research Council
  • Royal Academy of Engineering
  • Imperial College London
  • Marie Skłodowska-Curie Innovative Training Networks

Zitieren

APA

Ng, W. K., Dranczewski, J., Fischer, A., Raziman, T. V., Saxena, D., Farchy, T., Stenning, K., Peters, J., Schmid, H., Branford, W. R., Barahona, M., Moselund, K., Sapienza, R., & Gartside, J. C. (2026). Few-shot neuromorphic vision in a nonlinear photonic network laser. Science Advances. https://doi.org/10.1126/sciadv.aec6546

Vancouver

Ng WK, Dranczewski J, Fischer A, Raziman TV, Saxena D, Farchy T, et al. Few-shot neuromorphic vision in a nonlinear photonic network laser. Science Advances. 2026. doi:10.1126/sciadv.aec6546

BibTeX
@article{Ng2026,
  author = {Ng, Wai Kit and Dranczewski, Jakub and Fischer, Anna and Raziman, T V and Saxena, Dhruv and Farchy, Tobias and Stenning, Kilian and Peters, Jonathan and Schmid, Heinz and Branford, Will R and Barahona, Mauricio and Moselund, Kirsten and Sapienza, Riccardo and Gartside, Jack C},
  title = {Few-shot neuromorphic vision in a nonlinear photonic network laser},
  journal = {Science Advances},
  year = {2026},
  doi = {10.1126/sciadv.aec6546}
}
Kostenlos starten →