AI Paper Atlas

Research map / Optimization / Optimizers

Markov Chain Score Ascent: A Unifying Framework of Variational Inference with Markovian Gradients

Kyurae Kim, Jisu Oh, Jacob R. Gardner, Adji Bousso Dieng, Hongseok Kim

NeurIPS 2022 · Optimization › Optimizers · approach: Stochastic gradient MCMC · 3 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 pMCSA instead of ELBO N=10mean-field BNN · BNN regression, UCI energytest log predictive density: -2.40 -> -1.92Improved
use pMCSA instead of ELBO N=10mean-field BNN · BNN regression, UCI airfoiltest log predictive density: -2.56 -> -2.27Improved
use pMCSA instead of ELBO N=10mean-field BNN · BNN regression, UCI gastest log predictive density: 0.13 -> -0.09Hurt
use pMCSA instead of ELBO N=10mean-field BNN · BNN regression, UCI yachttest log predictive density: -2.44 -> -2.49No clear effect
use pMCSA instead of unlabeled MCSA variant 2mean-field BNN · BNN regression, UCI smltest log predictive density: -1.97 -> -1.22Improved
use pMCSA instead of unlabeled MCSA variant 1mean-field BNN · BNN regression, UCI gastest log predictive density: -0.47 -> -0.09Improved
use pMCSA instead of unlabeled MCSA variant 1mean-field BNN · BNN regression, UCI energytest log predictive density: -2.38 -> -1.92Improved
use pMCSA instead of unlabeled MCSA variant 1mean-field BNN · BNN regression, UCI winetest log predictive density: -0.97 -> -0.95No clear effect
increase ELBO budget from N=1 to N=10ELBO · BNN regression, UCI energytest log predictive density: -2.42 -> -2.40No clear effect
increase ELBO budget from N=1 to N=10ELBO · BNN regression, UCI gastest log predictive density: -0.06 -> 0.13Improved
increase computational budget N (2^3 to 2^7) for pMCSApMCSA · 50-D Gaussian, gradient variancegradient varianceImproved
increase computational budget N for MSCMSC · 50-D Gaussian, gradient variancegradient varianceNo 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