Transformer-based models to deal with heterogeneous environments in Human Activity Recognition

被引:4
|
作者
Ek S. [1 ]
Portet F. [1 ]
Lalanda P. [1 ]
机构
[1] Univ Grenoble Alpes, CNRS, Grenoble INP, Grenoble
关键词
Data heterogeneity; Human Activity Recognition; Machine learning; Transformers;
D O I
10.1007/s00779-023-01776-3
中图分类号
学科分类号
摘要
Human Activity Recognition (HAR) on mobile devices has been demonstrated to be possible using neural models trained on data collected from the device’s inertial measurement units. These models have used convolutional neural networks (CNNs), long short-term memory (LSTMs), transformers, or a combination of these to achieve state-of-the-art results with real-time performance. However, these approaches have not been extensively evaluated in real-world situations where the input data may be different from the training data. This paper highlights the issue of data heterogeneity in machine learning applications and how it can hinder their deployment in pervasive settings. To address this problem, we propose and publicly release the code of two sensor-wise transformer architectures called HART and MobileHART for Human Activity Recognition Transformer. Our experiments on several publicly available datasets show that these HART architectures outperform previous architectures with fewer floating point operations and parameters than conventional transformers. The results also show they are more robust to changes in mobile position or device brand and hence better suited for the heterogeneous environments encountered in real-life settings. Finally, the source code has been made publicly available. © 2023, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
引用
收藏
页码:2267 / 2280
页数:13
相关论文
共 50 条
  • [1] Transformer-based Models for Arabic Online Handwriting Recognition
    Alwajih, Fakhraddin
    Badr, Eman
    Abdou, Sherif
    INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS, 2022, 13 (05) : 898 - 905
  • [2] Classification and recognition of gesture EEG signals with Transformer-Based models
    Qu, Yan
    Li, Congsheng
    Jiang, Haoyu
    2024 3RD INTERNATIONAL CONFERENCE ON ROBOTICS, ARTIFICIAL INTELLIGENCE AND INTELLIGENT CONTROL, RAIIC 2024, 2024, : 415 - 418
  • [3] Transformer-based models for intrapulse modulation recognition of radar waveforms
    Bhatti, Sidra Ghayour
    Taj, Imtiaz Ahmad
    Ullah, Mohsin
    Bhatti, Aamer Iqbal
    ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2024, 136
  • [4] A transformer-based real-time earthquake detection framework in heterogeneous environments
    Wu, Aming
    Khan, Irshad
    Kwon, Young-Woo
    SCIENTIFIC REPORTS, 2025, 15 (01):
  • [5] Transformer-based Human Action Recognition with Dynamic Feature Selection
    Lamghari, Soufiane
    Bilodeau, Guillaume-Alexandre
    Saunier, Nicolas
    2023 20TH CONFERENCE ON ROBOTS AND VISION, CRV, 2023, : 129 - 136
  • [6] Transformer-based language-independent gender recognition in noisy audio environments
    Or Haim Anidjar
    Roi Yozevitch
    Scientific Reports, 15 (1)
  • [7] Human activity recognition and fall detection using convolutional neural network and transformer-based architecture
    Al-qaness, Mohammed A. A.
    Dahou, Abdelghani
    Abd Elaziz, Mohamed
    Helmi, Ahmed M.
    BIOMEDICAL SIGNAL PROCESSING AND CONTROL, 2024, 95
  • [8] Transformer-based deep reverse attention network for multi-sensory human activity recognition
    Pramanik, Rishav
    Sikdar, Ritodeep
    Sarkar, Ram
    ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2023, 122
  • [9] A transformer-based network for speech recognition
    Tang L.
    International Journal of Speech Technology, 2023, 26 (02) : 531 - 539
  • [10] Point transformer-based human activity recognition using high-dimensional radar point clouds
    Guo, Zhongyuan
    Guendel, Ronny G.
    Yarovoy, Alexander
    Fioranelli, Francesco
    2023 IEEE RADAR CONFERENCE, RADARCONF23, 2023,