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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

After Skin Wounding, Noncoding dsRNA Coordinates Prostaglandins and Wnts to Promote Regeneration.

The Journal of investigative dermatology·2017
Same author

Metabolic profiling and novel plasma biomarkers for predicting survival in epithelial ovarian cancer.

Oncotarget·2017
Same author

Identification of a six-lncRNA signature associated with recurrence of ovarian cancer.

Scientific reports·2017
Same author

Ropivacaine versus levobupivacaine in peripheral nerve block: A PRISMA-compliant meta-analysis of randomized controlled trials.

Medicine·2017
Same author

Fabrication of fluorescent composite hydrogel using in situ synthesis of upconversion nanoparticles.

Nanotechnology·2017
Same author

Ammonium assimilation: An important accessory during aerobic denitrification of Pseudomonas stutzeri T13.

Bioresource technology·2017

Related Experiment Video

Updated: Jul 11, 2025

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

3.8K

MAG-Res2Net: a novel deep learning network for human activity recognition.

Hanyu Liu1, Boyang Zhao1, Chubo Dai1

  • 1College of Medicine and Biological Information Engineering, Northeastern University, Shenyang 110819, People's Republic of China.

Physiological Measurement
|November 8, 2023
PubMed
Summary

This study introduces MAG-Res2Net, a novel deep learning model for human activity recognition (HAR). The model enhances feature extraction and achieves superior performance on benchmark datasets, improving HAR accuracy and efficiency.

Keywords:
attention mechanismborderline-SMOTEhuman activity recognitionmetric learningmulti-scale convolutional neural network

More Related Videos

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.4K
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.2K

Related Experiment Videos

Last Updated: Jul 11, 2025

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

3.8K
A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.4K
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.2K

Area of Science:

  • Computer Science
  • Biomedical Engineering
  • Machine Learning

Background:

  • Human Activity Recognition (HAR) is crucial for healthcare, sports, and fitness applications.
  • Current deep learning HAR methods struggle with diverse activities and data quality, hindering feature extraction.
  • Challenges in HAR include activity variability and data imperfections, impacting model accuracy.

Purpose of the Study:

  • To develop an advanced neural network model, MAG-Res2Net, for improved human activity recognition.
  • To address limitations in feature extraction caused by diverse activities and data quality issues in HAR.
  • To enhance HAR model performance and training efficiency through novel algorithmic integrations.

Main Methods:

  • Proposed MAG-Res2Net model integrating Borderline-SMOTE, metric learning-based loss combination, and Lion optimization.
  • Evaluation on UCI-HAR and WISDM public datasets for performance benchmarking.
  • Comparison with state-of-the-art models using the CSL-SHARE multimodal dataset.

Main Results:

  • Achieved 94.44% accuracy, 94.38% F1-macro, and 94.26% F1-weighted on UCI-HAR.
  • Attained 98.32% accuracy, 97.26% F1-macro, and 98.42% F1-weighted on WISDM.
  • Demonstrated robust multimodal performance and superior evaluation metrics and training efficiency over existing HAR networks.

Conclusions:

  • MAG-Res2Net exhibits strong multimodal HAR capabilities, with individual modules enhancing overall performance.
  • The proposed model outperforms current HAR neural networks in both accuracy and training efficiency.
  • The MAG-Res2Net model offers a significant advancement in human activity recognition technology.