Research map / Probabilistic and scientific modeling / Bayesian optimization

BO-Arena: An Evolving Benchmark for High-Dimensional Bayesian Optimisation

NeurIPS 2026 · Probabilistic and scientific modeling › Bayesian optimization · approach: Acquisition function design

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Argues that current high-dimensional Bayesian optimisation benchmarking practices are systematically ineffective and proposes BO-Arena, an evolving benchmark for comparing HDBO methods.

Problem

Benchmarking in high-dimensional Bayesian optimisation is ineffective, making it hard to determine which method is state-of-the-art as the field grows crowded.

Approach

An evolving benchmark (BO-Arena) intended to replace ad hoc per-paper evaluations with a continually updated, standardized comparison of HDBO methods.

Surprising findings

• Multiple published HDBO works fail to benchmark adequately, according to the abstract's claim that benchmarking practices in the field are systematically ineffective.

Closest papers by meaning

• Vanilla Bayesian Optimization Performs Great in High Dimensions (ICML 2024 · similarity 0.91)
• Understanding High-Dimensional Bayesian Optimization (ICML 2025 · similarity 0.91)
• BOIDS: High-Dimensional Bayesian Optimization via Incumbent-Guided Direction Lines and Subspace Embeddings (AAAI 2025 · similarity 0.90)
• Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal (NeurIPS 2024 · similarity 0.89)
• Posterior Inference with Diffusion Models for High-dimensional Black-box Optimization (ICML 2025 · similarity 0.89)
• Standard Gaussian Process is All You Need for High-Dimensional Bayesian Optimization (ICLR 2025 · similarity 0.88)