Nullifying the Inherent Bias of Non-invariant Exploratory Landscape Analysis Features

R.P. Prager, H. Trautmann, in: J. Correia, S. Smith, R. Qaddoura (Eds.), Applications of Evolutionary Computation, Springer Nature Switzerland, Cham, 2023, pp. 411–425.

Download
No fulltext has been uploaded.
Conference Paper | English
Author
Prager, Raphael Patrick; Trautmann, HeikeLibreCat
Editor
Correia, João; Smith, Stephen; Qaddoura, Raneem
Abstract
Exploratory landscape analysis (ELA) in single-objective black-box optimization relies on a comprehensive and large set of numerical features characterizing problem instances. Those foster problem understanding and serve as basis for constructing automated algorithm selection models choosing the best suited algorithm for a problem at hand based on the aforementioned features computed prior to optimization. This work specifically points to the sensitivity of a substantial proportion of these features to absolute objective values, i.e., we observe a lack of shift and scale invariance. We show that this unfortunately induces bias within automated algorithm selection models, an overfitting to specific benchmark problem sets used for training and thereby hinders generalization capabilities to unseen problems. We tackle these issues by presenting an appropriate objective normalization to be used prior to ELA feature computation and empirically illustrate the respective effectiveness focusing on the BBOB benchmark set.
Publishing Year
Proceedings Title
Applications of Evolutionary Computation
Page
411–425
LibreCat-ID

Cite this

Prager RP, Trautmann H. Nullifying the Inherent Bias of Non-invariant Exploratory Landscape Analysis Features. In: Correia J, Smith S, Qaddoura R, eds. Applications of Evolutionary Computation. Springer Nature Switzerland; 2023:411–425.
Prager, R. P., & Trautmann, H. (2023). Nullifying the Inherent Bias of Non-invariant Exploratory Landscape Analysis Features. In J. Correia, S. Smith, & R. Qaddoura (Eds.), Applications of Evolutionary Computation (pp. 411–425). Springer Nature Switzerland.
@inproceedings{Prager_Trautmann_2023, place={Cham}, title={Nullifying the Inherent Bias of Non-invariant Exploratory Landscape Analysis Features}, booktitle={Applications of Evolutionary Computation}, publisher={Springer Nature Switzerland}, author={Prager, Raphael Patrick and Trautmann, Heike}, editor={Correia, João and Smith, Stephen and Qaddoura, Raneem}, year={2023}, pages={411–425} }
Prager, Raphael Patrick, and Heike Trautmann. “Nullifying the Inherent Bias of Non-Invariant Exploratory Landscape Analysis Features.” In Applications of Evolutionary Computation, edited by João Correia, Stephen Smith, and Raneem Qaddoura, 411–425. Cham: Springer Nature Switzerland, 2023.
R. P. Prager and H. Trautmann, “Nullifying the Inherent Bias of Non-invariant Exploratory Landscape Analysis Features,” in Applications of Evolutionary Computation, 2023, pp. 411–425.
Prager, Raphael Patrick, and Heike Trautmann. “Nullifying the Inherent Bias of Non-Invariant Exploratory Landscape Analysis Features.” Applications of Evolutionary Computation, edited by João Correia et al., Springer Nature Switzerland, 2023, pp. 411–425.

Export

Marked Publications

Open Data LibreCat

Search this title in

Google Scholar
ISBN Search