Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Apr 17, 2026

Fabrication and Testing of Microfluidic Optomechanical Oscillators
09:10

Fabrication and Testing of Microfluidic Optomechanical Oscillators

Published on: May 29, 2014

12.8K

Programmable optical activation functions for KANs via feed-forward integrated microring resonators.

Hao Jia, Shanglin Yang

    Optics Letters
    |April 15, 2026
    PubMed
    Summary

    Researchers developed a programmable optical activation function unit using silicon photonics. This device offers tunable, diverse activation functions for optical neural networks (ONNs), enhancing flexibility and enabling new architectures like Kolmogorov-Arnold Networks (KANs).

    Related Concept Videos

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Previously unreported monomeric sorbicillinoids from Trichoderma reesei SCNU-F0042 as α-glucosidase inhibitors and antioxidants: isolation, bioactivity, and potential mechanisms.

    Phytochemistry·2026
    Same author

    DDIT3, OTUB2, and ASS1 regulate arginine biosynthesis in colorectal cancer cells under arginine deficiency.

    Cell reports·2026
    Same author

    Plant-derived extracellular vesicles as bioactive nanoplatforms: Biogenesis, therapeutic evidence, and translational challenges.

    Biochemistry and biophysics reports·2026
    Same author

    Differentiating high-frequency and high-severity hotspots: A robust risk-evolution-volume (REV) framework.

    Accident; analysis and prevention·2026
    Same author

    The metabolic change trajectory of skeletal muscle during the life cycle: arteriovenous metabolomics.

    Life medicine·2026
    Same author

    Deep learning-based quantitative histopathology of endoscopic biopsies in Crohn's disease: a retrospective cross-sectional validation study.

    Frontiers in immunology·2026

    Area of Science:

    • Photonics and Optical Engineering
    • Artificial Intelligence and Machine Learning
    • Materials Science for Computing

    Background:

    • Activation functions are crucial for optical neural networks (ONNs), but current methods demand high optical power or lack tunability.
    • Limited tunability restricts the design of versatile ONN architectures and the implementation of advanced models.

    Purpose of the Study:

    • To propose and demonstrate a novel programmable activation function unit for ONNs.
    • To achieve diverse and tunable activation functions using a silicon photonics platform.
    • To explore the application of this unit in advanced neural network architectures like Kolmogorov-Arnold Networks (KANs).

    Main Methods:

    • A feed-forward microring resonator with a tunable coupling coefficient was designed and fabricated on a silicon photonics platform.

    More Related Videos

    Fabrication and Characterization of High-Q Silicon Nitride Membrane Resonators
    09:46

    Fabrication and Characterization of High-Q Silicon Nitride Membrane Resonators

    Published on: August 8, 2025

    1.4K
    Microwave Photonics Systems Based on Whispering-gallery-mode Resonators
    12:18

    Microwave Photonics Systems Based on Whispering-gallery-mode Resonators

    Published on: August 5, 2013

    17.7K

    Related Experiment Videos

    Last Updated: Apr 17, 2026

    Fabrication and Testing of Microfluidic Optomechanical Oscillators
    09:10

    Fabrication and Testing of Microfluidic Optomechanical Oscillators

    Published on: May 29, 2014

    12.8K
    Fabrication and Characterization of High-Q Silicon Nitride Membrane Resonators
    09:46

    Fabrication and Characterization of High-Q Silicon Nitride Membrane Resonators

    Published on: August 8, 2025

    1.4K
    Microwave Photonics Systems Based on Whispering-gallery-mode Resonators
    12:18

    Microwave Photonics Systems Based on Whispering-gallery-mode Resonators

    Published on: August 5, 2013

    17.7K
  • The device's ability to generate diverse activation function profiles (monotonic and non-monotonic) was modeled and experimentally validated.
  • A physics-aware KAN framework was constructed utilizing the device's tunable solution space.

    • Main Results:

    • The proposed unit demonstrated flexible generation of a wide range of activation functions.
    • Experimental validation confirmed the device's tunable capabilities.
    • A KAN achieved 98.1% classification accuracy on the MNIST dataset when integrated with the device.

    Conclusions:

    • The developed programmable activation function unit offers significant design freedom for ONNs.
    • This technology paves the way for the monolithic integration of versatile and powerful optical neural networks.
    • The device's tunability is key for implementing complex models like KANs in optical hardware.