Research map / Large language models / Agents, code and math
Evolution of Benchmark: Black-Box Optimization Benchmark Design through Large Language Model
Chen Wang, Sijie Ma, Zeyuan Ma, Yue-Jiao Gong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train GLHF on EoB-UAV instead of CoCo-BBOB | GLHF · UAV test | final test performance: 8.380E-01 -> 8.484E-01 | Improved |
| train GLHF on EoB-HPO instead of CoCo-BBOB | GLHF · HPO test | final test performance: 5.144E-01 -> 6.580E-01 | Improved |
| train on EoB-BBOB instead of CoCo-BBOB | GLHF · UAV test | final test performance: 8.380E-01 -> 7.508E-01 | Hurt |
| use task-targeted EoB benchmark instead of CoCo-BBOB for consistency | EoB · UAV | performance consistency: 0.003 -> 0.453 | Improved |
| use task-targeted EoB benchmark instead of CoCo-BBOB for consistency | EoB · HPO | performance consistency: 0.694 -> 0.761 | Improved |
| use EoB proxy instead of real task for MetaBBO training step | LDE · UAV (normal) | wall-clock time: 30.38 min -> 3.124 s | Improved |
| use NeurELA instead of DeepELA for LSI features (ablation) | EoB · HPO, GLEET | final test performance: 0.7923 -> 0.8342 | Improved |
| use NeurELA instead of ELA Meta-Model features (ablation) | EoB · HPO, RLDEAFL | final test performance: 0.7891 -> 0.8593 | Improved |
| remove landscape construction knowledge from initialization prompt (ablation) | EoB · UAV-train Pareto front | LSI | Hurt |
| remove reflection stage (ablation) | EoB · UAV-train Pareto front | search efficiency | Hurt |
| increase neighborhood size to 10 and PBI theta to 10 (ablation) | EoB · UAV-train Pareto front | LSI | Mixed |
| reduce neighborhood size to 1 and theta to 1 (ablation) | EoB · UAV-train Pareto front | efficiency | Hurt |
Closest papers by meaning
- BBOWP-Bench: Evaluating LLMs on Black-Box Optimization Word Problems (NeurIPS 2026 · similarity 0.88)
- MetaBox-v2: A Unified Benchmark Platform for Meta-Black-Box Optimization (NeurIPS 2025 · similarity 0.86)
- Interpreting Neural Combinatorial Optimization via Evolving Programmatic Bottlenecks (NeurIPS 2026 · similarity 0.86)
- HSEvo: Elevating Automatic Heuristic Design with Diversity-Driven Harmony Search and Genetic Algorithm Using LLMs (AAAI 2025 · similarity 0.85)
- Simple Baselines are Competitive with Code Evolution (NeurIPS 2026 · similarity 0.85)
- DesignX: Human-Competitive Algorithm Designer for Black-Box Optimization (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
