AI Paper Atlas

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

A Tale of Two Latent Flows: Learning Latent Space Normalizing Flow with Short-Run Langevin Flow for Approximate Inference

Jianwen Xie, Yaxuan Zhu, Yifei Xu, Dingcheng Li, Ping Li

AAAI 2023 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Normalizing flow training and design · 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 latent normalizing-flow prior with short-run Langevin inference instead of LEBMLFBM image model · SVHNFID: 29.44 -> 23.64Improved
Use latent normalizing-flow prior with short-run Langevin inference instead of LEBMLFBM image model · CIFAR-10MSE: 0.020 -> 0.016Improved
Use latent normalizing-flow prior with short-run Langevin inference instead of LEBMLFBM image model · CelebAFID: 37.87 -> 33.64Improved
Replace learned inference network with short-run Langevin inference (ablation)LFBM-VAE · CelebA inpaintingMSE: 0.2074 -> 0.1937Improved
Replace learned inference network with short-run Langevin inference (ablation)LFBM-VAE · CelebA inpaintingFID: 0.0562 -> 0.0557Improved
Use latent normalizing-flow prior instead of EBM priorLatent prior model · CelebA unsupervised recovery, salt-and-pepper mask 30%MSE: 0.0256 -> 0.0244Improved
Use latent normalizing-flow prior instead of Gaussian priorLatent prior model · CelebA unsupervised recovery, salt-and-pepper mask 30%FID: 153.01 -> 46.2Improved
Use latent normalizing-flow prior instead of Gaussian priorLatent prior model · CelebA unsupervised recovery, single region mask 20 × 20MSE: 0.0404 -> 0.0420Hurt
Increase the depth of the flow-based prior (ablation)LFBM · CIFAR-10FIDImproved
Use MCMC inference instead of amortized inference (ablation)LFBM · CelebAMask-HQ 256 × 256FID: 129.65 -> 75.20Improved
Use MCMC inference instead of amortized inference (ablation)LFBM · CelebAMask-HQ 256 × 256training time: 5.5 -> 20.7Hurt
Use a normalizing-flow prior rather than an EBM priorLFBM · CelebA synthesissynthesis time: 116 -> 105Improved

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