thermal-dosimetry-surrogate

The repository contains the code for the paper Standardized Benchmark Dataset for Localized Exposure to a Realistic Source at 10-90 GHz.

This repository is related to the following conference papers:

• Standardized Benchmark Dataset for Localized Exposure to a Realistic Source at 10-90 GHz, in proceedings of BioEM 2023
• Prediction of Maximum Temperature Rise on Skin Surface for Local Exposure at 10-90 GHz, in proceedings of URSI GASS 2023

To reproduce the results, easiest way is to create a local environment by using conda as

conda create --name thermal-dosimetry-surrogate python=3.9.12

and, inside the environment, within code repository, run the following command

pip install -r requirements.txt

to install all dependencies listed in requirements.txt.

Contents

Directory	Subdirectory/Contents	Description
data
1	raw	Collected from the annex of the IEEE Std 2889-2021.
2	processed	Clean version of the collected data and the synthetic data set.
3	models	Results regarding the predictive performance of surrogate models.
figures		Generated figures and further augmented figures for conference papers, posters and presentations.
models		Parameters of fitted surrogate models.
notebooks		Jupyter notebooks.
1	00_data_processing.ipynb	Cleaning raw data, initial visualizations.
2	01_synthetic_data_generation.ipynb	Generating the synthetic data set.
3	02_baseline_model.ipynb	XGBoost baseline surrogate model.
4	03_multilayer_perceptron.ipynb	Feedforward neural network surrogate model.
5	04_mixture_of_experts.ipynb	Quadratic polynomial + tensor product splines.
6	05_tabnet.ipynb	TabNet surrogate model.
7	06_postprocessing.ipynb	Visualization of the predictive performance of surrogate models.
src		Code used in 03_multilayer_perceptron.ipynb notebook.

Cite

@misc{Kapetanovic2023Standardized,
      title={Standardized Benchmark Dataset for Localized Exposure to a Realistic Source at 10$-$90 {GHz}}, 
      author={Kapetanović, Ante and Poljak, Dragan and Li, Kun},
      year={2023},
      eprint={2305.02260},
      archivePrefix={arXiv},
      primaryClass={physics.med-ph},
      doi={10.48550/arXiv.2305.02260}
}

or

@inproceedings{Kapetanovic2023Prediction,
      title={Prediction of Maximum Temperature Rise on Skin Surface for Local Exposure at 10$-$90 {GHz}},
      author={Kapetanović, Ante and Poljak, Dragan and Li, Kun},
      year={2023},
      booktitle={2023 XXXVth General Assembly and Scientific Symposium of the International Union of Radio Science (URSI GASS)},
      doi={10.23919/URSIGASS57860.2023.10265331}}

License

MIT

Author

Ante Kapetanović