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

SGNet: Spectral-geometric neural network for structured representation learning.

Idowu Paul Okuwobi1, Jingyuan Liu2, Olayinka Susan Raji2

  • 1School of Life & Environmental Sciences, Guilin University of Electronic Technology, Guilin, 541004, China; Guangxi Academy of Artificial Intelligence, Nanning, Guangxi, 530201, China; Nantong Hamadun Medical Technology Co., Ltd, Nantong, Jiangsu, 226400, China.

Neural Networks : the Official Journal of the International Neural Network Society
|May 24, 2026
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

State Space Representation01:27

State Space Representation

The frequency-domain technique, commonly used in analyzing and designing feedback control systems, is effective for linear, time-invariant systems. However, it falls short when dealing with nonlinear, time-varying, and multiple-input multiple-output systems. The time-domain or state-space approach addresses these limitations by utilizing state variables to construct simultaneous, first-order differential equations, known as state equations, for an nth-order system.
Consider an RLC circuit, a...
Sequence Networks of Rotating Machines01:24

Sequence Networks of Rotating Machines

A Y-connected synchronous generator, grounded through a neutral impedance, is designed to produce balanced internal phase voltages with only positive-sequence components. The generator's sequence networks include a source voltage that is exclusively in the positive-sequence network. The sequence components of line-to-ground voltages at the generator terminals illustrate this configuration.
Zero-sequence current induces a voltage drop across the generator's neutral impedance and other...

You might also read

Related Articles

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

Sort by
Same author

A monoclonal antibody targeting OXA23 restores carbapenem susceptibility in global epidemic carbapenem-resistant <i>Acinetobacter baumannii</i>.

Antimicrobial agents and chemotherapy·2026
Same author

Accelerating Reaction Discovery through AI-HTE Integration: Nitrene-Mediated C-O Coupling as a Validated Case Study.

JACS Au·2026
Same author

The oral microbiome is associated with the diagnosis, prognosis and radiotherapy sensitivity of esophageal cancer.

Journal of translational medicine·2026
Same author

Optimal ultrasound enhances T. reesei cellulases secretion for high biomass enzymatic saccharification by hyphal fragmentation.

Ultrasonics sonochemistry·2026
Same author

A high-power plantar energy harvester with LSTM-KAN adaptive damping control.

iScience·2026
Same author

Albumin-corrected anion gap and 28-/90-day mortality in cirrhotic patients with invasive fungal infection: a retrospective cohort study.

BMC infectious diseases·2026
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

IGFD-Net: Illumination-guided frequency decoupling for polarization image fusion.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Multiple-Strategies dung beetle optimizer and its applications in engineering optimization and bankruptcy prediction.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Aggregating global-scale pixel-wise forgery cues within a graph.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Finite-Time intermittent control for secure synchronization of Neutral-Type stochastic delayed neural networks under aperiodic DoS attacks.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles

SGNet introduces a novel neural architecture that unifies spatial and spectral feature learning for improved computer vision tasks. This shape-aware approach achieves state-of-the-art results in image and 3D object recognition.

Area of Science:

  • Computer Vision
  • Deep Learning
  • Geometric Deep Learning

Background:

  • Convolutional Neural Networks (CNNs) have fixed local receptive fields, limiting global shape awareness.
  • Vision Transformers (ViTs) treat inputs as unordered sets, discarding crucial geometric structure.
  • Existing methods struggle to balance local spatial details with global shape context.

Purpose of the Study:

  • To propose SGNet, a novel neural architecture operating in a joint spectral-geometric representation space.
  • To develop a Spectral-Geometric Operator (SGO) that learns data-adaptive spectral bases.
  • To enable shape-aware frequency reasoning without handcrafted transforms or quadratic-complexity attention.

Main Methods:

  • SGNet learns features simultaneously in the local spatial domain and a data-adaptive spectral basis.
Keywords:
Convolutional neural networksManifold learningNeural operatorsSpectral-geometric learningVision transformers

Related Experiment Videos

  • The Spectral-Geometric Operator (SGO) constructs a geometric affinity graph and learns spectral bases.
  • A learnable gating mechanism fuses local geometric responses with global spectral coefficients.

    • Main Results:

    • SGNet achieves state-of-the-art performance on image classification (CIFAR-100, ImageNet-1K) and 3D object recognition (ModelNet40).
    • The architecture outperforms existing CNNs, ViTs, and hybrid models.
    • SGNet demonstrates superior performance with fewer parameters and FLOPs.

    Conclusions:

    • SGNet establishes a new direction in neural representation learning via adaptive operators on data-induced manifolds.
    • The proposed Spectral-Geometric Operator (SGO) offers theoretical stability and expressivity under geometric perturbations.
    • SGNet provides a powerful, efficient, and shape-aware framework for computer vision tasks.