Research map / Large language models / AI and society
Human-AI Collaborative Bayesian Optimisation
Arun Kumar Anjanapura Venkatesh, Santu Rana, Alistair Shilton, Svetha Venkatesh
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 human-AI teaming with recommendation correction (HAT-RR) | standard Bayesian optimisation · SVM tuning, Credit | test classification error (%): 18.1 -> 12.7 | Improved |
| add human-AI teaming with recommendation correction (HAT-RR) | standard Bayesian optimisation · SVM tuning, Ionosphere | test classification error (%): 8.9 -> 5.15 | Improved |
| add human-AI teaming with recommendation correction (HAT-RR) | standard Bayesian optimisation · SVM tuning, Car | test classification error (%): 1.9 -> 0.21 | Improved |
| add human-AI teaming with recommendation correction (HAT-RR) | standard Bayesian optimisation · SVM tuning, WDBC | test classification error (%): 1.5 -> 0.98 | Improved |
| use KFO-tuned ground-truth kernel instead of SE kernel in HAT-RR (ablation) | HAT-RR · SVM tuning, WDBC | test classification error (%): 0.98 -> 0.60 | Improved |
| use KFO-tuned ground-truth kernel instead of SE kernel in HAT-RR (ablation) | HAT-RR · SVM tuning, Biodeg | test classification error (%): 13.4 -> 13.88 | Hurt |
| use KFO-tuned ground-truth kernel instead of SE kernel in HAT-RR (ablation) | HAT-RR · SVM tuning, Ionosphere | test classification error (%): 5.15 -> 5.25 | Hurt |
| use KFO-tuned kernel in HAT-RR on synthetic functions (ablation) | HAT-RR · synthetic benchmarks (Ackley, Branin, Hartmann etc.) | simple regret | Improved |
| use expert-correction (HAT-RR) instead of good/bad-region feedback (HAT-DM) (ablation) | HAT framework with KFO · SVM tuning, Biodeg | test classification error (%): 14.09 -> 13.88 | Improved |
| naively add expert suggestions to observation model (STD-BO MOD) vs HAT-RR (ablation) | standard BO · SVM tuning, Credit | test classification error (%): 14.52 -> 12.7 | Improved |
| add noise (variance 0.1) to expert kernel weights and trade-off factor (ablation) | HAT-RR and HAT-DM · Levy 2D, Shubert 2D, Egg 2D | simple regret | No clear effect |
| human expert intervenes at every third iteration with ground-truth kernel | GP-UCB BO emulating expert · Oscillator 2D | kernel lengthscale learning | Improved |
Closest papers by meaning
- Principled Bayesian Optimization in Collaboration with Human Experts (NeurIPS 2024 · similarity 0.88)
- A No Free Lunch Theorem for Human-AI Collaboration (AAAI 2025 · similarity 0.87)
- Bayesian Optimization from Human Feedback: Near-Optimal Regret Bounds (ICML 2025 · similarity 0.87)
- Human Expertise in Algorithmic Prediction (NeurIPS 2024 · similarity 0.87)
- Beyond Single Stationary Policies: Meta-Task Players as Naturally Superior Collaborators (NeurIPS 2024 · similarity 0.85)
- Dynamic Knowledge Injection for AIXI Agents (AAAI 2024 · 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.
