An Imporved Masking Strategy for Self-supervised Masked Reconstruction in Human Activity Recognition

This repository is designed to implement the idea of "An Imporved Masking Strategy for Self-supervised Masked Reconstruction in Human Activity Recognition" in this paper.

Requirements

This project code is done in Python 3.8 and third party libraries.

TensorFlow 2.x is used as a deep learning framework.

The main third-party libraries used and the corresponding versions are as follows:

• tensorflow 2.3.1

• tensorflow_addons 0.15.0

• numpy 1.18.5

• scikit-learn 0.23.1

Running

This demo can be run with the following command:

python main.py

Code Organisation

The main content of each file is marked as follows:

• dataset.py: Dataset processing
• encoder.py: Model structure
• encoderLayer.py: Model structure
• main.py: Main Program Entry
• module.py: Components needed to train the model
• multiHeadAttention.py: Model structure
• utils.py: The tools and methods needed to train the model

Citation

If you find our paper useful or use the code available in this repository in your research, please consider citing our work:

@article{wang2023improved,
  title={An Improved Masking Strategy for Self-supervised Masked Reconstruction in Human Activity Recognition},
  author={Wang, Jinqiang and Zhu, Tao and Ning, Huansheng},
  journal={arXiv preprint arXiv:2312.04147},
  year={2023}
}

Reference

• https://github.com/dapowan/LIMU-BERT-Public