Sabri Manai, Szymon Bobek, Mehrdad Asadi, Ann Nowe, Grzegorz J. Nalepa
Hyperparameter optimization (HPO) is usually framed as a fully automated search that returns a single “best” configuration, leaving human operators with little insight into alternatives or trade-offs. However, in many applied settings, practitioners arrive with concrete wishes such as higher precision, less labeling effort, or faster models that standard HPO tools cannot address directly. In this work, we propose a surrogate-based explainable AI (XAI) framework that turns informal requirements into counterfactual queries over the configuration space. A CatBoost model is trained on the resulting configuration-performance log and analysed with XAI tools, then queried by the DiCE and CFNOW explainers to generate counterfactual configurations under fixed soft constraints. Experiments on four standard binary classification benchmarks show that the framework can provide locally reliable guidance. In particular, DiCE often suggests alternative settings that result in consistent improvements when the main model is retrained.
Manai, S, Bobek, S, Asadi, M, Nowe, A & J. Nalepa, G 2026, Counterfactual User Guidance for Improving Transparent Hyperparameter Tuning. in M Paszynski, AS Barnard & YJ Zhang (eds), Computational Science – ICCS 2026 Workshops: 26th International Conference, ICCS 2026, Hamburg, Germany, June 29 – July 1, 2026, Proceedings, Part IV. Lecture Notes in Computer Science, vol. 16789 LNCS, Springer, pp. 262-276, 26th International Conference on Computational Science, Hamburg, Germany, 29/06/26. https://doi.org/10.1007/978-3-032-29918-5_19
Manai, S., Bobek, S., Asadi, M., Nowe, A., & J. Nalepa, G. (2026). Counterfactual User Guidance for Improving Transparent Hyperparameter Tuning. In M. Paszynski, A. S. Barnard, & Y. J. Zhang (Eds.), Computational Science – ICCS 2026 Workshops: 26th International Conference, ICCS 2026, Hamburg, Germany, June 29 – July 1, 2026, Proceedings, Part IV (pp. 262-276). (Lecture Notes in Computer Science; Vol. 16789 LNCS). Springer. https://doi.org/10.1007/978-3-032-29918-5_19
@inproceedings{e390dac7c1e9415da9b56cde3a9ce929,
title = "Counterfactual User Guidance for Improving Transparent Hyperparameter Tuning",
abstract = "Hyperparameter optimization (HPO) is usually framed as a fully automated search that returns a single “best” configuration, leaving human operators with little insight into alternatives or trade-offs. However, in many applied settings, practitioners arrive with concrete wishes such as higher precision, less labeling effort, or faster models that standard HPO tools cannot address directly. In this work, we propose a surrogate-based explainable AI (XAI) framework that turns informal requirements into counterfactual queries over the configuration space. A CatBoost model is trained on the resulting configuration-performance log and analysed with XAI tools, then queried by the DiCE and CFNOW explainers to generate counterfactual configurations under fixed soft constraints. Experiments on four standard binary classification benchmarks show that the framework can provide locally reliable guidance. In particular, DiCE often suggests alternative settings that result in consistent improvements when the main model is retrained.",
author = "Sabri Manai and Szymon Bobek and Mehrdad Asadi and Ann Nowe and \{J. Nalepa\}, Grzegorz",
note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 26th International Conference on Computational Science, ICCS ; Conference date: 29-06-2026 Through 01-07-2026",
year = "2026",
month = jun,
day = "27",
doi = "10.1007/978-3-032-29918-5\_19",
language = "English",
isbn = "9783032299178",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "262--276",
editor = "Maciej Paszynski and Barnard, \{Amanda S.\} and Zhang, \{Yongjie Jessica\}",
booktitle = "Computational Science – ICCS 2026 Workshops",
url = "https://www.iccs-meeting.org/iccs2026/",
}