NeuroEVOlution through Geometric Semantic perturbation and Population-based Training (NEVO-GSPT) is a population-based neural network evolution method, that adapts the inflate and deflate geometric semantic mutations of GSGP and SLIM to NeuroEvolution.
You can find an example usage script in example.py.
The paper, accepted and presented at EvoStar2026, can be found here.
If you use NEVO-GSPT in a scientific publication, please consider citing the following paper:
@misc{farinati2026nevogsptpopulationbasedneuralnetwork,
title={NEVO-GSPT: Population-Based Neural Network Evolution Using Inflate and Deflate Operators},
author={Davide Farinati and Frederico J. J. B. Santos and Leonardo Vanneschi and Mauro Castelli},
year={2026},
eprint={2601.08657},
archivePrefix={arXiv},
primaryClass={cs.NE},
url={https://arxiv.org/abs/2601.08657},
}