On the Structural (Dis)Agreement of Landscape Representations in Black-Box Optimization

2026-05-27

AI Summary

This study systematically investigates the consistency and divergence among four state-of-the-art landscape representation methods—ELA, DeepELA, TransOptAS, and DoE2Vec—in characterizing the structural properties of problem instances from the MA-BBOB benchmark suite within black-box optimization. Through unsupervised clustering, coverage stability assessment, cross-representation similarity analysis, and empirical performance evaluation using differential evolution and particle swarm optimization, the work reveals significant discrepancies and complementary strengths among the representations: ELA and TransOptAS yield compact structures, DeepELA exhibits balanced behavior, and DoE2Vec demonstrates strong semantic alignment albeit with fragmentation. The study introduces a multi-perspective landscape analysis framework, demonstrating that no single representation suffices for comprehensive algorithm performance prediction, thereby establishing a new paradigm for landscape-informed algorithm selection.

Abstract

Landscape feature representations play a central role in automated algorithm selection and meta-learning for black-box optimization, yet little is known about how different representations agree (or disagree) in the structures they impose on problem spaces. This paper presents a systematic unsupervised evaluation of four state-of-the-art representations (ELA, DeepELA, TransOptAS, and DoE2Vec) using a diverse set of affine combinations of BBOB functions (MA-BBOB). By applying extensive clustering analyses, coverage-based stability measures, and cross-representation similarity assessments, we show that each representation organizes the same problems in markedly different ways: ELA and TransOptAS form compact geometric structures, DeepELA provides a balanced intermediate view, and DoE2Vec achieves strong semantic alignment but with substantial fragmentation. Our results reveal that no single representation dominates; rather, they capture complementary aspects of the underlying landscapes. These findings highlight the importance of multi-view analyses for understanding representation behavior and offer guidance on selecting or combining representations in downstream meta-learning and algorithm selection tasks. In addition, across two different algorithm families (Differential Evolution and Particle Swarm Optimization), we show that landscape representations face an inherent trade-off in how well they align structural landscape descriptions with observed performance, indicating that no single representation can fully capture algorithm performance.

Problem

• landscape representation
• black-box optimization
• structural disagreement
• algorithm selection
• meta-learning

Innovation

• landscape representation
• black-box optimization
• algorithm selection
• unsupervised evaluation
• meta-learning