Code for article: Misaligned Over-The-Air Computation of Multi-Sensor Data with Wiener-Denoiser Network
Use jupyter notebook and run src/main.ipynb. Follow the notebook for dataset synthesis, training and evaluating.
It is recommend to run it on kaggle, and use Gas sensor array temperature modulation dataset as input. Otherwise, you have to download the csv data and modify paths in src/gen_dataset.py.
Our wiener estimator is also implemented into Over-the-air Federate Learning, for the fast and accurate misaligned gradient aggression, please refer to https://github.com/auto-Dog/Fed_MisAligned_OAC