New unsupervised multi-kernel learning method for automated algorithm selection

Researchers have developed a novel unsupervised multi-kernel learning approach for automated algorithm selection in black-box optimization. This method groups problem instances based on heterogeneous landscape representations without relying on performance labels during the clustering phase. The approach utilizes a multi-kernel k-means formulation to learn cluster assignments and kernel weights across four different landscape views: ELA, DeepELA, DoE2Vec, and TransOptAS. Evaluations on differential evolution and particle swarm optimization tasks demonstrated that this unsupervised method achieves strong performance, particularly for differential evolution, and remains competitive with leading baselines for particle swarm optimization. AI

IMPACT This unsupervised approach could reduce the cost and improve the generalization of algorithm selection in optimization tasks.

RANK_REASON The item is an academic paper detailing a new methodology in machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• BBOB
• DeepELA
• differential evolution
• DoE2Vec
• particle swarm optimization
• TransOptAS: Transformer-Based Algorithm Selection for Single-Objective Optimization
• Unsupervised Multi-kernel Learning for Automated Algorithm Selection

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →