AI Paper Atlas

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

Deterministic Langevin Monte Carlo with Normalizing Flows for Bayesian Inference

Richard D.P. Grumitt, Biwei Dai, Uros Seljak

NeurIPS 2022 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Normalizing flow training and design · 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
remove Metropolis-Hastings correction (ablation)DLMC · hierarchical funnel d=101, sigma=5posterior qualityHurt
remove normalizing-flow preconditioning (ablation)DLMC · sparse logistic regression d=51convergence speedHurt
remove Langevin step, use NF-proposal MH only (ablation)DLMC · funnel d=101, sigma=0.1convergence speedHurt
replace normalizing flow with kernel density estimation (PP-DLMC) (ablation)DLMC · benchmark examplesposterior qualityHurt
replace DLMC with SVGD (ablation)particle methods · benchmark examplesposterior qualityHurt
use parallel instead of serial DLMCDLMC · hierarchical funnel d=101, sigma=5wall-clock time: 30 minutesImproved
use DLMC instead of NUTSsampler · hierarchical funnel d=101, sigma=5likelihood evaluations: 15,000 likelihood callsImproved
parallel DLMC+HMC instead of NUTSsampler · ill-conditioned Gaussian d=1000wall-clock evaluations: 10^5 -> 100Improved
serial DLMC+HMC versus NUTSsampler · ill-conditioned Gaussian d=1000serial likelihood evaluations: 10^5No clear effect
DLMC alone without HMCDLMC · ill-conditioned Gaussian d=1000biasHurt
upsample to 1000 instead of 5000 particlesDLMC · sparse logistic regression d=51computational costNo clear effect
S-DLMC versus NUTSsampler · Rosenbrock 32-dwall-clock qualityNo clear effect

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