BBOB Function

BBOB functions are a widely used benchmark suite for evaluating the performance of black-box optimization algorithms, primarily focusing on continuous, single-objective problems. Recent research emphasizes generating new benchmark instances through affine combinations of existing BBOB functions (MA-BBOB), allowing for a more nuanced exploration of algorithm behavior across diverse problem landscapes and facilitating the development and testing of automated algorithm selection methods. This work investigates how landscape features and algorithm performance relate, aiming to improve the efficiency of selecting training sets for machine learning models used in algorithm selection. The insights gained from this research contribute to a more robust and comprehensive understanding of algorithm performance and ultimately improve the design and application of optimization algorithms in various fields.