Few-Shot Learning in Screening Septuple-Atomic-Layer Supported Single-Atom Catalysts for Hydrogen Production

Overview

This repository contains the implementation of the machine learning approach described in our paper "Few-shot learning for screening 2D Ga2CoS4- x supported single-atom catalysts for hydrogen production" for predicting Hydrogen Evolution Reaction (HER) activity. The model uses a combination of convolutional and dense neural networks to predict HER activity based on material properties.

Citation

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

@article{khossossi2025few,
  title={Few-shot learning for screening 2D Ga2CoS4- x supported single-atom catalysts for hydrogen production},
  author={Khossossi, Nabil and Dey, Poulumi},
  journal={Journal of Energy Chemistry},
  volume={100},
  pages={665--673},
  year={2025},
  publisher={Elsevier}
}

Requirements

• Python 3.8+
• PyTorch 1.8+
• NumPy 1.19+
• Matplotlib 3.3+
• SciPy 1.6+
• xlrd 2.0+

Input Data Format

The model expects an Excel file with the following columns:

1. Electronegativity
2. d-orbital of metal
3. Group number
4. Radius (pm)
5. First ionization energy
6. HER activity (target variable)

Usage

1. Clone the repository:

git clone https://github.com/NabKh/HER_SACs_JEC2024.git
cd HER_SACs_JEC2024

2. Install requirements:

pip install -r requirements.txt

3. Prepare your data:

   • Place your Excel file in data/
   • Ensure it follows the format described above

4. Run the training:

python src/main.py

5. Results will be saved in results/run_YYYYMMDD_HHMMSS/:
   • best_model.pkl: Trained model state
   • training_losses.txt: Training and test losses
   • training_errors.txt: Training and test errors
   • training_history.png: Learning curves plot
   • train_predictions.txt: Training set predictions
   • test_predictions.txt: Test set predictions
   • new_predictions.txt: Predictions for new elements

Contributing

We welcome contributions and the use of the script in more large and different datasets! Please feel free to submit a Pull Request.

Contact

• Dr. Nabil Khossossi
• Email: [email protected]