PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
May 17, 2026IET BiometricsOpen Access

Multiscale Convolutional–Bidirectional LSTM Fusion of Spatiotemporal Attention for Human Activity Recognition

View Full Paper
Ask AI
Bookmark
Share

Authors

YZYuyang ZhangCXChangcheng XiangYZYinrui Zhang

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates improved activity recognition accuracy in smart healthcare technologies, suggesting better real-world application.

Key Points

  • The aim is to enhance human activity recognition through an innovative model that integrates multiple technologies.
  • Developed a model combining multiscale convolution, BiLSTM, and spatiotemporal attention.
  • Utilized filter sizes of 3, 5, 7, and 9 in a multiscale parallel convolutional structure.
  • Employed the Swish activation function to optimize feature extraction and address gradient issues.
  • Achieved an accuracy of 95.39% on the UCI-HAR public dataset.
  • Demonstrated excellent performance of 99.58% on a custom dataset.
  • Significantly improved accuracy through enhanced feature selection and noise resilience.

Cite This Study

Zhang et al. (2026) studied this question.

synapsesocial.com/papers/6a095c2c7880e6d24efe229chttps://doi.org/10.1049/bme2/8129974
View Full Paper
Ask AI
Bookmark
Share