AI Paper Atlas

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

Projected Latent Markov Chain Monte Carlo: Conditional Sampling of Normalizing Flows

Chris Cannella, Mohammadreza Soltani, Vahid Tarokh

ICLR 2021 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Stochastic gradient MCMC · 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
Average 10 conditional samples instead of using an individual sample (ablation)PL-MCMC MC-EM imputation · MNIST, multiple missingness mechanisms and ratesReconstruction RMSE: 0.153(1) -> 0.130(2) (I.M. 0.3); 0.261(2) -> 0.256(1) (I.M. 0.9)Improved
Average conditional samples rather than use an individual sample (ablation)PL-MCMC test-set imputation · MNIST, multiple missingness mechanisms and ratesFID: 1.56(7) -> 1.58(8) (I.M. 0.3); 87(2) -> 90(2) (I.M. 0.9)Mixed
Use MC-EM trained PL-MCMC flows rather than MisGANMNIST imputation · MNIST I.M. 0.9FID: 90(2) vs 11(1)Hurt
Use MC-EM trained PL-MCMC flows rather than MisGANMNIST imputation · MNIST P.M. 0.9Reconstruction RMSE: 0.2648(9) vs 0.3693(9)Improved
Use MC-EM trained PL-MCMC flows rather than MIWAEUCI imputation · UCI datasets with 50% uniform missingnessNMSE: banknote: 0.58(3) vs 0.56(4); yeast: 0.96(6) vs 0.95(5)Mixed
Use latent-space proposals instead of Gibbs sampling in modeled data spacePL-MCMC conditional sampling · MNIST conditional expectationRMSE versus proposal numberImproved
Use a more restrictive auxiliary distribution (ablation)PL-MCMC conditional sampling · MNIST conditional expectationRMSE versus proposal numberMixed
Increase the transition proposal scale (ablation)PL-MCMC conditional sampling · MNIST conditional expectationRMSE versus proposal numberMixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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