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: Jan 6, 2026

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

982

Building a Compact Convolutional Neural Network for Embedded Intelligent Sensor Systems Using Group Sparsity and

Jungchan Cho1, Minsik Lee2

  • 1Department of Software, Gachon University, Seongnam 13120, Korea. thinkai@gachon.ac.kr.

Sensors (Basel, Switzerland)
|October 9, 2019
PubMed
Summary

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

Age-stratified analysis of descending aorta diameter in traumatic massive hemorrhage: a machine learning approach.

Trauma surgery & acute care open·2026
Same author

Deep Learning-Based Aortic Diameter Measurement in Traumatic Hemorrhage Using Shallow Attention Network: A Path Forward.

Diagnostics (Basel, Switzerland)·2025
Same author

Text-guided cross-position attention for image analysis: Case of medical image.

Computers in biology and medicine·2025
Same author

Shallow and reverse attention network for colon polyp segmentation.

Scientific reports·2023
Same author

Deep-Learning-Based Stroke Screening Using Skeleton Data from Neurological Examination Videos.

Journal of personalized medicine·2022
Same author

Occluded Pedestrian-Attribute Recognition for Video Sensors Using Group Sparsity.

Sensors (Basel, Switzerland)·2022

This study introduces two strategies for creating efficient deep learning networks for intelligent sensors. A novel feedback control mechanism optimizes the trade-off between network size and performance, improving efficiency without sacrificing accuracy.

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Machine Learning

Background:

  • Growing popularity of AI and deep learning necessitates efficient models for real-time applications.
  • Embedding deep networks in resource-constrained intelligent sensors faces challenges, particularly the accuracy-efficiency trade-off.
  • Existing methods struggle to balance network complexity reduction with performance maintenance.

Purpose of the Study:

  • To propose novel strategies for designing compact deep networks for intelligent sensors.
  • To maintain high performance while minimizing computational resources and processing time.
  • To address the trade-off between network sparsity and accuracy.

Main Methods:

  • Developed two strategies for compact deep network design.
Keywords:
convolutional neutral networkdeep learninggroup sparsityknowledge distillationparameter reduction.

Related Experiment Videos

Last Updated: Jan 6, 2026

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

982
  • First strategy utilizes group sparsity and knowledge distillation (KD) for automatic parameter determination.
  • Second strategy employs a feedback control mechanism based on proportional control theory to balance sparsity and accuracy.

    • Main Results:

    • The proposed methods effectively create compact deep networks with minimal performance degradation.
    • The feedback control mechanism optimizes the trade-off curve between network size and accuracy.
    • Experiments on CIFAR-10, CIFAR-100, and ImageNet datasets demonstrate superior performance compared to baselines.

    Conclusions:

    • The proposed strategies enable the development of efficient deep learning models for intelligent sensors.
    • The feedback control mechanism successfully navigates the accuracy-sparsity trade-off, improving overall efficiency.
    • This research offers a promising approach for deploying advanced AI in resource-limited edge devices.