Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Contextual Causal Bayesian Optimisation
Vahan Arsenyan, Antoine Grosnit, Haitham Bou Ammar, Arnak S. Dalalyan
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 |
|---|---|---|---|
| Optimise over mixed policy scopes (CoCa-BO) instead of causal-only CaBO | CaBO · PSA SCM, config I (CaBO can reach optimum) | normalised cumulative regret | Improved |
| Condition on context (CoCa-BO) instead of CaBO | CaBO · PSA SCM, config II (context needed) | normalised cumulative regret | Improved |
| Use CoCa-BO instead of CoBO when both can reach optimum | CoBO · SCM config I | normalised cumulative regret | Hurt |
| Use CoCa-BO instead of CoBO when fixed scope is suboptimal | CoBO · SCM config II | normalised cumulative regret | Improved |
| Use CoCa-BO instead of CoBO with many variables | CoBO · large-variable SCM | sample complexity: dimensionality reduced by factor of 22 | Improved |
| Increase noise std in Y up to 1.25 (ablation) | CoCa-BO · noise robustness | normalised regret: sigma 0.01 to 1.25 | No clear effect |
| Extreme noise in Y (sigma 6.25) (ablation) | CoCa-BO · noise robustness | normalised regret: sigma = 6.25 | Hurt |
| Change GP backend (HEBO, BoTorch, GPflow) (ablation) | CoCa-BO · PSA-like SCM | regret | No clear effect |
| Remove contextual variables from all POMPS (ablation) | CoCa-BO · ablation environment | regret | Improved |
Closest papers by meaning
- Constrained Causal Bayesian Optimization (ICML 2023 · similarity 0.91)
- Differentiable Multi-Target Causal Bayesian Experimental Design (ICML 2023 · similarity 0.90)
- Adversarial Causal Bayesian Optimization (ICLR 2024 · similarity 0.89)
- Exploiting Causal Graph Priors with Posterior Sampling for Reinforcement Learning (ICLR 2024 · similarity 0.89)
- Intervention and Conditioning in Causal Bayesian Networks (NeurIPS 2024 · similarity 0.88)
- Partial Structure Discovery is Sufficient for No-regret Learning in Causal Bandits (NeurIPS 2024 · 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.
