AI Paper Atlas

Research map / Probabilistic and scientific modeling / Bayesian optimization

Distributionally Robust Bayesian Optimization with $\varphi$-divergences

Hisham Husain, Vu Nguyen, Anton van den Hengel

NeurIPS 2023 · Probabilistic and scientific modeling › Bayesian optimization · approach: Acquisition function design · 1 citations (OpenAlex)

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 changedWhereMeasuredEffect
Use large epsilon_t in {0.5,1} when robust and stochastic optima differ (ablation)DRBO · Benchmark functions, robust optimum differs from stochastic optimumcumulative robust regretImproved
Set epsilon_t toward 0 when robust and stochastic optima coincide (ablation)DRBO · Benchmark functions, identical optimaoptimization performanceImproved
Use adaptive epsilon_t instead of a fixed epsilon (ablation)DRBO · Benchmark functionsoptimization performance stabilityImproved
Replace MMD adversary (linear programming) with closed-form phi-divergence reductionDRBO · 5d benchmark functioncomputational cost per iteration: 5-times fasterImproved
Replace MMD adversary with closed-form phi-divergence reductionDRBO · 6d benchmark functioncomputational cost per iteration: 10-times fasterImproved
Use chi2/TV DRBO instead of DRBO MMDDRBO · Benchmark functionscumulative robust regretImproved
Use distributional robustness instead of vanilla BO ignoring contextBO · Benchmark functionscumulative robust regretImproved
Use distributional robustness instead of Stable-Opt worst-case over predefined contextsStable-Opt · Benchmark functionscumulative robust regretImproved
Increase context set size |C| from 30 to 100 (ablation)DRBO chi2/TV/KL · Wind Poweroptimization performance: 30 -> 100Improved
Increase context set size |C| from 100 to 500 (ablation)DRBO chi2/TV/KL · Wind Poweroptimization performance: 100 -> 500No clear effect
Increase context set size |C| (ablation)DRBO MMD · Wind Poweroptimization performance and costMixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy