AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

Information-Theoretic Safe Exploration with Gaussian Processes

Alessandro Giacomo Bottero, Carlos E. Luis, Julia Vinogradska, Felix Berkenkamp, Jan Peters

NeurIPS 2022 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: PAC-Bayes and decoupled GP learning · 3 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 a too small or too large Lipschitz constant L in StageOpt exploration (ablation)StageOpt · 1D f(x)=exp(-x)+0.05percentage of domain classified safeHurt
Replace StageOpt (Lipschitz-based expanders) with ISE information-gain acquisitionsafe exploration · 2D GP samples (100 samples, 100 iterations)percentage of domain classified safeImproved
Compare ISE with the heuristic SafeOpt-inspired acquisitionISE · 2D GP samplespercentage of domain classified safeNo clear effect
Use ISE instead of StageOptsafe exploration · 2D GP samplespercentage of safety violations per runNo clear effect
Use ISE instead of StageOptsafe exploration · Cart pole controller (3D)percentage of true safe set classified safeNo clear effect
Use LineBO version of ISE exploiting observation noise instead of StageOpt-like explorationISE · 9D and 10D heteroskedastic noisesimple regretImproved

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