Official pytorch implementation of "ConvBoost: Boosting ConvNets for Sensor-based Activity Recognition". (accepted at Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2023)
This work proposes ConvBoost – a novel, three layer, structured model architecture and boosting framework for convolutional network based HAR. Our framework generates additional training data from three different perspectives for improved HAR, aiming to alleviate the shortness of labeled training data in the field
This code provides an implementation for ConvBoost. This repository is implemented using PyTorch and it includes code for running experiments on OPP, Pamap2, GOTOV datasets.
To ensure the experiments run smoothly, please create a python 3.8 environment.
you could run a baseline model by:
python main.py --dataset=1 --model='CNN-3L'
To run the ConvBoost model by:
python main.py --dataset=1 --model='CNN-3L' --epoch_wise=True
More configuration settings can be found in the file 'cfg_data.py'
If you found this paper is helpful and like it. Please don't mind citing it and thank you.
@article{10.1145/3596234
author = {Shao, Shuai and Guan, Yu and Zhai, Bing and Missier, Paolo and Pl\"{o}tz, Thomas},
title = {ConvBoost: Boosting ConvNets for Sensor-based Activity Recognition},
doi = {10.1145/3596234},
journal = {Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.},
volume = {7},
number = {2},
year = {2023},
issue_date = {June 2023},
pages = {75},
url = {https://doi.org/10.1145/3596234},
publisher = {Association for Computing Machinery},
}