Prof. Bartz-Beielstein’s talk „Surrogate Model-Based Multi-Objective Optimization Using Desirability
Functions“ was presented at ACM Genetic and Evolutionary Computation Conference in Malaga.
Abstract:
The desirability function approach is an established and widely adopted method in industry for optimizing multiple response processes. It is seldomly used in multi-criteria hyperparameter tuning. This article fills this gap. It provides an introduction to the desirability function approach to multi-objective optimization (direct and surrogate model-based), and multi-objective hyperparameter tuning.
This work is based on the paper by Kuhn (2026).
It presents a Python implementation of Kuhn’s R package „desirability“. After the desirability-function approach is introduced, two examples are given that demonstrate how to use desirability
functions for classical optimization via response surface modeling and hyperparameter tuning of a neural network, which is implemented in PyTorch.
The article discusses the following research questions:
- How can the desirability function approach be used for multi-objective optimization and hyperparameter tuning?
- What are advantages and disadvantages of the desirability function approach compared to other multi-objective optimization methods?
- How can the desirability function approach be improved?
References:
- Kuhn, M. 2016. “Desirability: Function Optimization and Ranking via Desirability Functions. https://cran.r-project.org/package=desirability.
- Bartz-Beielstein, Thomas. 2025. “Multi-Objective Optimization and Hyperparameter Tuning With Desirability Functions.” arXiv e-Prints, March, arXiv:2503.23595. https://doi.org/10.48550/arXiv.2503.23595.
Slides and Notebook:
- Slides [PDF]: https://advm1.gm.fh-koeln.de/~bartz/Gecco2025/bart25b-beamer.pdf
- Jupyter Notebook: https://advm1.gm.fh-koeln.de/~bartz/Gecco2025/bart25b-beamer.ipynb