AI Paper Atlas

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

Stochastic Variational Inference With Gradient Linearization

Tobias Plötz, Anne S. Wannenwetsch, Stefan Roth

CVPR 2018 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Variational inference theory and gradients · 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
Use gradient linearization for stochastic variational inference instead of gradient-based SVI with AdamSVIGL · Sintel validation crops (100×100)unnormalized KL: 4.24 -> 3.78Improved
Use gradient linearization for stochastic variational inference instead of gradient-based SVI with SGDSVIGL · Sintel validation crops (100×100)unnormalized KL: 4.45 -> 3.78Improved
Use gradient linearization for stochastic variational inference instead of AdamSVIGL · BSDS test denoisingunnormalized KL: 0.98 -> 0.97No clear effect
Use gradient linearization for stochastic variational inference instead of AdamSVIGL · BSDS test denoisingPSNR: 24.70 -> 24.77Improved
Use gradient linearization for stochastic variational inference instead of AdamSVIGL · BSDS test denoisingSSIM: 0.680 -> 0.693Improved
Use SVIGL instead of MAP estimation with L-BFGSOptical flow estimation · Sintel testAEPE: 5.81 -> 5.74Improved
Vary the Adam step size (ablation)SVI with Adam · Optical flow and Poisson-Gaussian denoisingunnormalized KLMixed
Reduce the sample set size (ablation)SVI with Adam · Optical flow and Poisson-Gaussian denoisingunnormalized KLHurt
Reduce the sample set size (ablation)SVIGL · Optical flow and Poisson-Gaussian denoisingconvergenceNo clear effect
Use Gaussian mean-field variational inference with SVIGL instead of a Laplace approximationPoisson-Gaussian denoising · BSDS testunnormalized KL: 1.57 -> 0.97Improved

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