AI Paper Atlas

Research map / Probabilistic and scientific modeling / Bayesian optimization

Deep Bayesian Optimization on Attributed Graphs

Jiaxu Cui, Bo Yang, Xia Hu

AAAI 2019 · Probabilistic and scientific modeling › Bayesian optimization · approach: Multi-fidelity hyperparameter optimization · 6 citations (OpenAlex)

Read the paper

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 DGBO rather than GPs-based BO with graphlet kernelsSynthetics · Synthetics, situations (a)-(d)objective y: situation (a): 2.815\pm0.002 -> 2.863\pm0.000; situation (b): 2.815\pm0.002 -> 2.863\pm0.000; situation (c): 2.815\pm0.0Mixed
Use the proposed graph convolution rather than existing graph kernelsGraph Bayesian optimization · Syntheticsobjective yImproved
Include node and edge features instead of ignoring edge features (ablation)DGBO · Delaneystability of finding the optimumImproved
Add basis regularization to edge-feature convolution (ablation)DGBO_Rel · ZINCoptimization performanceImproved
Use global attributes unrelated to the objective (ablation)Graph optimization methods · Synthetics, situation (d)objective y: DGBO_noRel: 2.863\pm0.000; BO_vec: 2.851\pm0.000; BO_Glets: 2.850\pm0.000; BO_dGlets: 2.821\pm0.001; GBO_dGlets: 2.849\pHurt
Use DGBO rather than VAE+GPsMolecular optimization · ZINCevaluations to find optimum as percentage of search space: 1.8%Improved
Use DGBO rather than simulated annealingRoad network design · SiouxFallsevaluations to find optimum: less than 420 evaluations; SA: ~3250 evaluations; nearly 7.7 times fasterImproved

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