AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Scalable Model Selection for Belief Networks

Zhao Song, Yusuke Muraoka, Ryohei Fujimaki, Lawrence Carin

NeurIPS 2017 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Variational methods for sequence models · 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
add FABIA model selection (gFIC shrinkage) to NVIL, 1-layerNVIL · MNISTtest VLB (nats): -115.63 -> −114.96Improved
add FABIA model selection, 2-layerNVIL · MNISTtest VLB (nats): -105.62 -> −104.92Improved
add FABIA model selection, 3-layerNVIL · MNISTtest VLB (nats): -101.99 -> −101.14Improved
shrink model size via FABIA, 1-layerNVIL · MNISTtest time (seconds): 8.47 -> 6.88Improved
shrink model size via FABIA, 2-layerNVIL · MNISTtest time (seconds): 12.34 -> 9.18Improved
shrink model size via FABIA, 3-layerNVIL · MNISTtest time (seconds): 15.66 -> 10.97Improved
use FABIA instead of VB with beta-normal prior shrinkage, 1-layerVB · MNISTtest VLB (nats): -117.04 -> −114.96Improved
use FABIA instead of Rec-MCEM, 2-layerRec-MCEM · MNISTtest VLB (nats): -106.54 -> −104.92Improved
initialize NVIL with true model size instead of FABIA's initial sizeNVIL · synthetic one- and two-layer SBNtest log-likelihoodImproved
FABIA vs NVIL (True) with the true model sizeFABIA · synthetic one- and two-layer SBNtest log-likelihood convergenceNo clear effect
use FABIA instead of NVILNVIL · RCV1test perplexity: 857 -> 856No clear effect
use FABIA instead of NVILNVIL · Wikipediatest perplexity: 735 -> 730No clear effect

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