WISE-HAR: A Generalizable Ensemble Deep Learning Framework for WiFi-Based Human Activity Recognition
The paper presents WISE-HAR, an ensemble deep learning framework for recognizing human activities using WiFi signals. It addresses challenges such as performance variance and small dataset size through innovative techniques like ensemble learning and data augmentation. The results demonstrate high accuracy and strong generalization capabilities, making it suitable for real-world applications.
- ▪WISE-HAR recognizes three activities: 'No Presence', 'Walking', and 'Walking + Arm-waving'.
- ▪The ensemble model achieved a test accuracy of 94.87% on the Line-of-Sight scenario.
- ▪Data augmentation improved Random Forest performance from 60% to 95%.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2606.02974 (cs) [Submitted on 2 Jun 2026] Title:WISE-HAR: A Generalizable Ensemble Deep Learning Framework for WiFi-Based Human Activity Recognition Authors:Maheen Arshad, Qindeel E Zahra, Muhammad Khuram Shahzad View a PDF of the paper titled WISE-HAR: A Generalizable Ensemble Deep Learning Framework for WiFi-Based Human Activity Recognition, by Maheen Arshad and 2 other authors View PDF HTML (experimental) Abstract:Human Activity Recognition (HAR) using WiFi signals has emerged as a transformative technology for smart homes, healthcare monitoring, security systems, and ambient assisted living.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
