AI Paper Atlas

Research map / Probabilistic and scientific modeling / Bayesian optimization

Reinforced Few-Shot Acquisition Function Learning for Bayesian Optimization

Bing-Jing Hsieh, Ping-Chun Hsieh, Xi Liu

NeurIPS 2021 · Probabilistic and scientific modeling › Bayesian optimization · approach: Acquisition function design · 2 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 Bayesian DQN with 5 particles instead of 1 particle (ablation)FSAF · Optimization benchmark functionstesting simple regretImproved
Remove the demo replay buffer from the Bayesian DQN prior (ablation)FSAF · Training across GP taskstraining stabilityHurt
Replace the chaser meta-loss with TD loss (ablation)FSAF · Training across GP taskstraining progressHurt
Increase the number of few-shot gradient updates for small K (ablation)FSAF · Optimization benchmark functionssimple regretImproved
Use FSAF rather than fixed benchmark acquisition functionsBayesian optimization on standard benchmark functions · Ackley, Eggholder, Dixon-Price, Styblinski-Tang, and Powellsimple regretMixed
Adapt FSAF with few-shot task dataFSAF · Five real-world test functionssimple regretImproved
Use FSAF on unseen 5-dimensional GP functions rather than MetaBOFSAF · Synthetic GP, 5Dsimple regretImproved
Use FSAF on 5-dimensional Matérn-2/3 functions rather than EIFSAF · Synthetic GP, 5D Matérn-2/3simple regretHurt
Fine-tune MetaBO using meta-dataMetaBO · Standard optimization benchmark functionssimple regretMixed

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