EMForecaster

Overview

This is the official repository for the paper:

Xavier Mootoo, Hina Tabassum, Luca Chiaraviglio. EMForecaster: A Deep Learning Framework for Time Series Forecasting in Wireless Networks with Distribution-Free Uncertainty Quantification (IEEE Transactions on Network Science and Engineering, under review).

Description

EMForecaster is a novel deep learning architecture specialized for time series forecasting, benchmarked primarily on electromagnetic field (EMF) exposure forecasting. EMForecaster also includes a conformal prediction pipeline for uncertainty quantification, along with a trade-off score metric which we propose a unified measure of model performance to balance width of prediction intervals (minimizing) and empirical coverage (maximizing).

Installation

Dependencies

• Python $\geq$ 3.10

Using conda

# Create and activate conda environment
conda create -n emforecaster python=3.10
conda activate emforecaster

# Install requirements
pip install -e .

Using pip

# Create and activate virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install requirements
pip install -e .

Data

Data is proprietary and provided by primarily by Luca Chiaraviglio, please contact him for access.

Usage

Running Experiments

Execute the main script with the desired model:

python main.py <model>

Results are saved in the logs folder if using offline logging. To run custom experiments, run emforecaster/jobs/exp/<model>/args.yaml for each model, with respect to the classes in config.py.

Hyperparameter Tuning

Hyperparameter tuning is also available, by modifying emforecaster/jobs/exp/<model>/ablation.yaml, which performs grid search on a set of parameters. To run hyperparameter tuning jobs use:

python /tuning/tune.py <model>

Neptune Tracking

This project allows for online experimental logging via Neptune.ai, which is free for researchers and students.

1. Create a Neptune.ai account and API token
2. Set your Neptune API token as an environment variable:

export NEPTUNE_API_TOKEN='your-neptune-api-token'

3. Set neptune: True in your args.yaml under the exp category. See other parameters such as run_id and exp_id.

The get_results.py reads your <model>.yaml file in the emforecaster/analysis/neptune/ablations directory, which iterates through all model runs (e.g., from a hyperparameter run) to display the best result with respect to a deciding_metric (e.g., MSE). All model runs and their model configurations are displayed and saved as results.csv.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citations

If you use this code in your research or work, please cite our paper:

@article{mootoo2025emforecaster,
    title     = {EMForecaster: A Deep Learning Framework for Time Series Forecasting in Wireless Networks with Distribution-Free Uncertainty Quantification},
    author    = {Mootoo, Xavier and Chiaraviglio, Luca and Tabassum, Hina},
    year      = {2025},
}

Contact

For queries, please contact the corresponding author through: xmootoo at gmail dot com.

Acknowledgments

Xavier Mootoo is supported by Canada Graduate Scholarships - Master's (CGS-M) funded by the Natural Sciences and Engineering Research Council (NSERC) of Canada, the Vector Scholarship in Artificial Intelligence, provided through the Vector Institute, Canada, and the Ontario Graduate Scholarship (OGS) granted by the provincial government of Ontario, Canada.

We extend our gratitude to Commune AI for generously providing the computational resources needed to carry out our experiments, in particular, we thank Luca Vivona (@LVivona) and Sal Vivona (@salvivona).