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 Videos

Optical dual-scale architecture for neural image recognition.

M A Neifeld

    Applied Optics
    |November 10, 2010
    PubMed
    Summary

    A new dual-scale neural network architecture enhances image recognition by reducing training costs and improving noise tolerance. This novel approach offers superior generalization performance compared to conventional networks.

    Related Concept Videos

    Parallel Processing01:20

    Parallel Processing

    The brain processes sensory information rapidly due to parallel processing, which involves sending data across multiple neural pathways at the same time. This method allows the brain to manage various sensory qualities, such as shapes, colors, movements, and locations, all concurrently. For instance, when observing a forest landscape, the brain simultaneously processes the movement of leaves, the shapes of trees, the depth between them, and the various shades of green. This enables a quick and...
    Visual System01:26

    Visual System

    Light enters the eye through the cornea, a transparent, dome-shaped surface covering the surface of the eyeball that helps to direct and focus incoming light. This light is then channeled toward the pupil, an adjustable opening whose size is controlled by the iris. The iris, a pigmented muscle, regulates the amount of light entering the eye by contracting or dilating the pupil, thereby ensuring optimal light levels for clear vision.
    Once through the pupil, the light passes through the lens, a...

    You might also read

    Related Articles

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

    Sort by
    Same author

    Optical memory: introduction by the feature editors.

    Applied optics·2010
    Same author

    Optical design for page access to volume optical media.

    Applied optics·2010
    Same author

    Error-correction schemes for volume optical memories.

    Applied optics·2010
    Same author

    Adaptive, optical, radial basis function neural network for handwritten digit recognition.

    Applied optics·2010
    Same author

    Lens-design issues affecting parallel readout of optical disks.

    Applied optics·2010
    Same author

    Improvements in the capacity of computer-generated holographic storage using the Lee method with sparse multivalued reconstructions.

    Applied optics·2010

    Area of Science:

    • Computer Science
    • Artificial Intelligence
    • Machine Learning

    Background:

    • Conventional neural networks face challenges with high training costs and limited noise tolerance in image recognition.
    • Generalization performance is crucial for real-world image recognition tasks, especially with distorted or noisy data.

    Purpose of the Study:

    • To introduce a novel dual-scale neural-network architecture for image recognition.
    • To demonstrate reduced training costs, improved noise tolerance, and enhanced generalization performance.
    • To explore the optical implementation feasibility of the proposed architecture.

    Main Methods:

    • Development of a novel dual-scale neural-network architecture integrating data reduction and attention mechanisms.
    • Simulation-based evaluation of the dual-scale network against conventional counterparts.
    • Demonstration of an example optical system implementing the dual-scale architecture.
    • Application of the network to a human face recognition problem.

    Main Results:

    • Simulations showed a 6.7x reduction in training cost for the dual-scale network.
    • A 67.3% improvement in noise tolerance was observed compared to conventional networks.
    • Generalization to distortions improved by 61.6% in the best-case scenario.
    • The architecture proved amenable to optical implementation.

    Conclusions:

    • The dual-scale neural network offers significant advantages in training efficiency, noise robustness, and generalization for image recognition.
    • The architecture's suitability for optical implementation opens avenues for hardware acceleration.
    • The demonstrated improvements suggest potential for advanced applications, including human face recognition.

    Related Experiment Videos