AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

Structured Neural Networks for Density Estimation and Causal Inference

Asic Q Chen, Ruian Shi, Xiang Gao, Ricardo Baptista, Rahul G Krishnan

NeurIPS 2023 · Probabilistic and scientific modeling › Causal discovery · approach: Latent-variable causal estimation · 1 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 structure via StrNN to autoregressive flow (StrAF vs ARF)autoregressive flow · 15-d synthetictest NLL: -3.09 -> -3.55Improved
StrAF versus GNFStrAF · 15-d synthetictest NLL: -3.63 vs -3.55No clear effect
add StrNN structure to continuous flow (StrCNF vs FFJORD)CNF · 15-d synthetictest NLL: -1.85 -> -4.01Improved
StrCNF versus Weilbach structured CNFStrCNF · 15-d synthetictest NLL: -2.59 -> -4.01Improved
replace MADE masks with greedy StrNN masks (ablation)StrNN · Random Sparse n=5000test NLL: 7.790 -> 7.758Improved
replace MADE masks with greedy StrNN masks (ablation)StrNN · Previous 3 n=2000test NLL: 8.816 -> 8.768Improved
replace MADE masks with greedy StrNN masks (ablation)StrNN · Every Other n=2000test NLL: 8.364 -> 8.315Improved
exact integer programming instead of greedy factorization (ablation)StrNN · Random Sparse n=5000test NLL: 7.758 -> 7.758No clear effect
add variance penalty to IP objective (IP-var) (ablation)StrNN · Previous 3 n=5000test NLL: 8.755 -> 8.758No clear effect
Zuko masking instead of greedy (ablation)StrNN · Random Sparse n=2000test NLL: 7.778 -> 7.776No clear effect
greedy masks instead of Zuko masks on star graph (ablation)StrAF · Star-shaped d=1024test NLL: -14.550 -> -18.734Improved
greedy masks instead of Zuko masks on star graph (ablation)StrAF · Star-shaped d=50test NLL: 1.3276 -> 0.4311Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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