Research map / Learning theory and algorithms / Prediction and decision losses
Improving the Expected Improvement Algorithm
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace standard EI with TTEI using beta=1/2 | EI · [5, 4, 1, 1, 1], c=0.95 | average measurements to reach confidence: 238.50 -> 14.60 | Helped |
| Replace standard EI with TTEI using beta=1/2 | EI · [5, 4, 3, 2, 1], c=0.95 | average measurements to reach confidence: 384.73 -> 16.72 | Helped |
| Replace standard EI with TTEI using beta=1/2 | EI · [2, .8, .6, .4, .2], c=0.95 | average measurements to reach confidence: 1525.42 -> 24.39 | Helped |
| Use adaptive beta instead of fixed beta=1/2 (ablation) | TTEI · [5, 4, 1, 1, 1], c=0.9999 | average measurements to reach confidence: 61.97 -> 61.98 | No clear effect |
| Set beta to beta* instead of beta=1/2 (ablation) | TTEI · [5, 4, 1, 1, 1], c=0.9999 | average measurements to reach confidence: 61.97 -> 61.59 | Helped |
| Use adaptive beta instead of fixed beta=1/2 (ablation) | TTEI · [5, 4, 3, 2, 1], c=0.9999 | average measurements to reach confidence: 66.56 -> 65.54 | Helped |
| Set beta to beta* instead of beta=1/2 (ablation) | TTEI · [5, 4, 3, 2, 1], c=0.9999 | average measurements to reach confidence: 66.56 -> 65.55 | Helped |
| Use adaptive beta instead of fixed beta=1/2 (ablation) | TTEI · [2, .8, .6, .4, .2], c=0.9999 | average measurements to reach confidence: 76.21 -> 72.94 | Helped |
| Set beta to beta* instead of beta=1/2 (ablation) | TTEI · [2, .8, .6, .4, .2], c=0.9999 | average measurements to reach confidence: 76.21 -> 71.62 | Helped |
| Use TTEI with adaptive beta rather than tracking optimal proportions directly | Tracking oracle (TO) · [5, 4, 1, 1, 1], c=0.9999 | average measurements to reach confidence: 77.76 -> 61.98 | Helped |
| Use TTEI with adaptive beta rather than tracking optimal proportions directly | Tracking oracle (TO) · [5, 4, 3, 2, 1], c=0.9999 | average measurements to reach confidence: 88.02 -> 65.54 | Helped |
| Use TTEI with adaptive beta rather than tracking optimal proportions directly | Tracking oracle (TO) · [2, .8, .6, .4, .2], c=0.9999 | average measurements to reach confidence: 96.90 -> 72.94 | Helped |
Closest papers by meaning
- Convergence Rates of Constrained Expected Improvement (NeurIPS 2025 · similarity 0.86)
- Expected Hypervolume Improvement Is a Particular Hypervolume Improvement (AAAI 2025 · similarity 0.84)
- Fixing the Loose Brake: Exponential-Tailed Stopping Time in Best Arm Identification (ICML 2025 · similarity 0.83)
- Optimal Batched Best Arm Identification (NeurIPS 2024 · similarity 0.83)
- A Unified Framework for Entropy Search and Expected Improvement in Bayesian Optimization (ICML 2025 · similarity 0.83)
- Fixed Confidence Best Arm Identification in the Bayesian Setting (NeurIPS 2024 · similarity 0.83)
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.