Health prediction for king salmon via evolutionary machine learning with genetic programming

Fangfang Zhang, Yuye Zhang, Paula Casanovas, Jessica Schattschneider, Seumas P. Walker, Bing Xue, Mengjie Zhang, Jane E. SymondsJournal of the Royal Society of New Zealand 2024 · DOI

Abstract

King (Chinook) salmon is the only salmon species farmed in Aotearoa New Zealand and accounts for over half of the world's production of king salmon. Determining the health status of king salmon effectively is important for farming. However, it is a challenging task due to the complex biotic and abiotic factors that influence health. Evolutionary machine learning algorithms have shown their superiority in learning models for challenging tasks. However, they have not been investigated for health prediction in king salmon farming. This paper focuses on data processing and machine learning algorithm design to develop king salmon health prediction models in Aotearoa New Zealand. Particularly, this paper proposes a king salmon health prediction method based on genetic programming which is an evolutionary machine learning algorithm. The results show that genetic programming achieves the best overall performance among all examined typical machine learning algorithms for most trials. Further analyses show that genetic programming can automatically detect important features for learning classifiers for king salmon health classification tasks effectively, and can also learn potentially interpretable models. Our results are an important step forward in developing health prediction tools to automatically assess health status of farmed king salmon in Aotearoa New Zealand.
BibTeX
@article{zhang2024health,
  author = {Fangfang Zhang and Yuye Zhang and Paula Casanovas and Jessica Schattschneider and Seumas P. Walker and Bing Xue and Mengjie Zhang and Jane E. Symonds},
  title = {{Health prediction for king salmon via evolutionary machine learning with genetic programming}},
  journal = {Journal of the Royal Society of New Zealand},
  year = {2024},
  doi = {10.1080/03036758.2024.2329228},
}

Loading related content...

Figure