Research map / Probabilistic and scientific modeling / Bayesian optimization
DABO: Difficulty-Aware Bayesian Optimization with Diffusion-Learned Priors
Mengyang Li, Pinlong Zhao
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 |
|---|---|---|---|
| add diffusion-generated training data to ifBO (ablation) | ifBO · LCBench (ablation) | final normalized regret: 0.016 -> 0.0152 | Improved |
| add difficulty conditioning in the PFN (ablation) | ifBO · LCBench (ablation) | final normalized regret: 0.016 -> 0.0148 | Improved |
| add difficulty-adaptive acquisition (ablation) | ifBO · LCBench (ablation) | final normalized regret: 0.016 -> 0.0154 | Improved |
| combine diffusion data with difficulty conditioning (ablation) | ifBO · LCBench (ablation) | final normalized regret: 0.016 -> 0.0142 | Improved |
| add all three difficulty-aware components (ablation) | ifBO · LCBench (ablation) | final normalized regret: 0.016 -> 0.0140 | Improved |
| difficulty-aware PFN trained on diffusion data vs FT-PFN | FT-PFN · PD1 surrogate | log-likelihood: 1.13 -> 2.50 | Improved |
| difficulty-aware PFN vs FT-PFN on MSE | FT-PFN · Taskset surrogate | MSE: 0.004 -> 0.0042 | No clear effect |
| difficulty-aware PFN vs FT-PFN inference time | FT-PFN · Surrogate runtime | runtime (sec): 0.72 -> 2.05 | Hurt |
| replace hand-crafted basis functions with diffusion-learned priors | ifBO · Synthetic curve fidelity | FCD: 2.83 -> 1.24 | Improved |
| DABO vs ifBO on PD1 final regret | ifBO · PD1 | final normalized regret: 0.034 -> 0.028 | Improved |
| DABO vs ifBO on easy tasks | ifBO · 75 HPO tasks | regret | No clear effect |
Closest papers by meaning
- Posterior Inference with Diffusion Models for High-dimensional Black-box Optimization (ICML 2025 · similarity 0.89)
- Efficient Hyperparameter Optimization with Adaptive Fidelity Identification (CVPR 2024 · similarity 0.88)
- Large Language Models to Enhance Bayesian Optimization (ICLR 2024 · similarity 0.87)
- Informed Initialization for Bayesian Optimization and Active Learning (NeurIPS 2025 · similarity 0.87)
- Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal (NeurIPS 2024 · similarity 0.87)
- Cost-Sensitive Freeze-thaw Bayesian Optimization for Efficient Hyperparameter Tuning (NeurIPS 2025 · similarity 0.87)
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.
