AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Online Structure Learning for Feed-Forward and Recurrent Sum-Product Networks

Agastya Kalra, Abdullah Rashwan, Wei-Shou Hsu, Pascal Poupart, Prashant Doshi, Georgios Trimponias

NeurIPS 2018 · Probabilistic and scientific modeling › Molecules and proteins · approach: Neuro-symbolic probabilistic layers · 13 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
oSLRAU structure learning instead of random networks of equal sizeSPN · Poweraverage log-likelihood: -2.83 -> -2.46Improved
oSLRAU structure learning versus random network structureSPN · Networkaverage log-likelihood: -5.34 -> -4.27Improved
periodic pruning of stale subtrees (ablation)oSLRAU · Powerlog-likelihood / size: -2.46 -> -2.40; nodes 23360 -> 5330Improved
periodic pruning of stale subtrees (ablation)oSLRAU · GasSenlog-likelihood: -102 -> -130Hurt
periodic pruning of stale subtrees (ablation)oSLRAU · MSDlog-likelihood: -527.7 -> -526.8No clear effect
periodic pruning on a nonstationary ordered MNIST stream (ablation)oSLRAU · MNIST ordered by labelgenerated samplesMixed
oSLRAU versus RealNVP onlineoSLRAU · MSDaverage log-likelihood: -531.4 vs -362.4Hurt
oSLRAU instead of search-and-score for recurrent SPNsRSPN · eegEyeaverage log-likelihood: 25.9 -> 36.9Improved
oSLRAU RSPN versus LSTM RNNRSPN+oSLRAU · ozLevelaverage log-likelihood: -30.1 vs -15.3Hurt

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