Datasets and Pytorch code for UniTS: Short-Time Fourier Inspired Neural Networks for Sensory Time Series Classification, in The 19th ACM Conference on Embedded Networked Sensor Systems (SenSys 2021).
PyTorch >= 1.8.0
numpy
scikit-learn
pytorch-complex
- Run main.py for UniTS and the following baseline models (ResNet, MaCNN, RFNet-base, THAT, LaxCat)
- Run complex_main.py for model STFNets.
https://drive.google.com/file/d/1aPb-iy6ic-bcg-azXVQ2_2uegn0oQc_j/view?usp=sharing
Motion:
Seizure:
https://physionet.org/content/chbmit/1.0.0/
WiFi:
https://github.com/ermongroup/Wifi_Activity_Recognition
KETI:
https://github.com/Shuheng-Li/Relational-Inference/tree/master/KETI_oneweek
Code will use cuda by default. You may tune model hyperparameter for better results.
Please cite the following paper if you use this repository in your research work:
@inproceedings{10.1145/3485730.3485942,
author = {Li, Shuheng and Chowdhury, Ranak Roy and Shang, Jingbo and Gupta, Rajesh K. and Hong, Dezhi},
title = {UniTS: Short-Time Fourier Inspired Neural Networks for Sensory Time Series Classification},
year = {2021},
isbn = {9781450390972},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3485730.3485942},
doi = {10.1145/3485730.3485942}
}
Contact Shuheng Li ✉️ for questions, comments and reporting bugs.