AI Paper Atlas

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

Decomposition for Bayesian Networks: Local and Parallel Inference

Pei Heng, Xinyi Hu, Yi Sun

TPAMI 2026 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Approximate inference in Bayesian neural networks · 0 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 minimal d-decomposition tree pruning instead of junction treejunction-tree inference · Pathfinder joint query, |Q|=2inference time: 0.7743 -> 0.0613Improved
use pruning instead of lossless decompositionlossless decomposition · Pathfinder joint query, |Q|=2inference time: 0.5061 -> 0.0613Improved
use pruning instead of junction tree at |Q|=10junction-tree inference · Pathfinder joint query, |Q|=10inference time: 0.7744 -> 0.1855Improved
use pruning instead of junction tree at |Q|=20junction-tree inference · Pathfinder joint query, |Q|=20inference time: 0.8450 -> 0.2966Improved
use pruning instead of junction tree at |Q|=25junction-tree inference · Pathfinder joint query, |Q|=25inference time: 1.8515 -> 1.8073No clear effect
use pruning instead of lossless decomposition at |Q|=25lossless decomposition · Pathfinder joint query, |Q|=25inference time: 30.1472 -> 1.8073Improved
increase query dimension from 2 to 25minimal d-decomposition tree inference · Pathfinder joint queryinference time: 0.0613 -> 1.8073Hurt
parallel parameter learning on decomposed sub-models instead of global learningglobal parameter estimation · Child (small network)speed-up ratio: 1.40Mixed
parallel parameter learning on decomposed sub-models instead of global learningglobal parameter estimation · Pigs (large network)speed-up ratio: 2.69Improved
increase sample size to 10,000parallel decomposition learning · Multinomial BNsKL, Hellinger and Bhattacharyya distanceImproved
increase sample size to 10,000parallel decomposition learning · Gaussian BNsWasserstein-2 distance: 0.04Improved

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