1 code implementation • 8 Jun 2023 • Gjorgjina Cenikj, Gašper Petelin, Carola Doerr, Peter Korošec, Tome Eftimov
The application of machine learning (ML) models to the analysis of optimization algorithms requires the representation of optimization problems using numerical features.
no code implementations • 1 Jun 2023 • Ana Nikolikj, Sašo Džeroski, Mario Andrés Muñoz, Carola Doerr, Peter Korošec, Tome Eftimov
In black-box optimization, it is essential to understand why an algorithm instance works on a set of problem instances while failing on others and provide explanations of its behavior.
no code implementations • 31 May 2023 • Ana Nikolikj, Gjorgjina Cenikj, Gordana Ispirova, Diederick Vermetten, Ryan Dieter Lang, Andries Petrus Engelbrecht, Carola Doerr, Peter Korošec, Tome Eftimov
A key component of automated algorithm selection and configuration, which in most cases are performed using supervised machine learning (ML) methods is a good-performing predictive model.
no code implementations • 30 May 2023 • Ana Nikolikj, Michal Pluháček, Carola Doerr, Peter Korošec, Tome Eftimov
That is, instead of considering cosine distance in the feature space, we consider a weighted distance measure, with weights depending on the relevance of the feature for the regression model.
no code implementations • 25 Apr 2022 • Gjorgjina Cenikj, Ryan Dieter Lang, Andries Petrus Engelbrecht, Carola Doerr, Peter Korošec, Tome Eftimov
Fair algorithm evaluation is conditioned on the existence of high-quality benchmark datasets that are non-redundant and are representative of typical optimization scenarios.
1 code implementation • 15 Apr 2022 • Ana Kostovska, Diederick Vermetten, Sašo Džeroski, Carola Doerr, Peter Korošec, Tome Eftimov
In addition, we have shown that by using classifiers that take the features relevance on the model accuracy, we are able to predict the status of individual modules in the CMA-ES configurations.
no code implementations • 22 Mar 2022 • Risto Trajanov, Stefan Dimeski, Martin Popovski, Peter Korošec, Tome Eftimov
Predicting the performance of an optimization algorithm on a new problem instance is crucial in order to select the most appropriate algorithm for solving that problem instance.
1 code implementation • 22 Oct 2021 • Risto Trajanov, Stefan Dimeski, Martin Popovski, Peter Korošec, Tome Eftimov
In this study, we are investigating explainable landscape-aware regression models where the contribution of each landscape feature to the prediction of the optimization algorithm performance is estimated on a global and local level.
no code implementations • 29 Sep 2021 • Tome Eftimov, Gašper Petelin, Gjorgjina Cenikj, Ana Kostovska, Gordana Ispirova, Peter Korošec, Jasmin Bogatinovski
By observing discrepancy between the empirical results of the bootstrap evaluation and recently adapted practices in TSC literature when introducing novel methods we warn on the potentially harmful effects of tuning the methods on certain parts of the landscape (unless this is an explicit and desired goal of the study).
no code implementations • 27 Apr 2021 • Urban Škvorc, Tome Eftimov, Peter Korošec
When designing a benchmark problem set, it is important to create a set of benchmark problems that are a good generalization of the set of all possible problems.
no code implementations • 22 Apr 2021 • Tome Eftimov, Anja Jankovic, Gorjan Popovski, Carola Doerr, Peter Korošec
Accurately predicting the performance of different optimization algorithms for previously unseen problem instances is crucial for high-performing algorithm selection and configuration techniques.