Transformer-Assisted Genetic Programming for Symbolic Regression

Codes for Paper Transformer-Assisted Genetic Programming for Symbolic Regression

PGGP

Pre-trained Model Guided Genetic Programming (PGGP) method is designed to assist the initialization and mutation of GP.

Getting Started

You should create a new folder 'weights' and download the Transformer weights from https://github.com/SymposiumOrganization/NeuralSymbolicRegressionThatScales/tree/main

After download the model, you can run PGGP by:

python main.py

Citation

@article{han2025transformer,
  title={Transformer-Assisted Genetic Programming for Symbolic Regression [Research Frontier]},
  author={Han, Xiaoxu and Zhong, Jinghui and Ma, Zhitong and Mu, Xin and Gligorovski, Nikola},
  journal={IEEE Computational Intelligence Magazine},
  volume={20},
  number={2},
  pages={58--79},
  year={2025},
  publisher={IEEE}
}