AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Flow Contrastive Estimation of Energy-Based Models

Ruiqi Gao, Erik Nijkamp, Diederik P. Kingma, Zhen Xu, Andrew M. Dai, Ying Nian Wu

CVPR 2020 · Label-efficient and robust learning › Semi-supervised learning · approach: Self-training theory · 60 citations (OpenAlex)

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
jointly train Glow and an EBM with the shared FCE objective instead of training Glow by MLEGlow · SVHNFID: 41.70 -> 20.19Improved
jointly train Glow and an EBM with the shared FCE objective instead of training Glow by MLEGlow · CIFAR-10FID: 45.99 -> 37.30Improved
jointly train Glow and an EBM with the shared FCE objective instead of training Glow by MLEGlow · CelebAFID: 23.32 -> 12.21Improved
use FCE rather than MLE to train GlowGlow · SVHN testbits per dimension: 2.17 -> 2.25Hurt
use FCE rather than MLE to train GlowGlow · CIFAR-10 testbits per dimension: 3.35 -> 3.45Hurt
use FCE rather than MLE to train GlowGlow · CelebA testbits per dimension: 3.49 -> 3.54Hurt
add joint FCE training after the EBM-only initialization stage (ablation)Conv-small EBM on SVHN · SVHN, 500 labelsclassification error (%): 9.42 -> 7.05Improved
add joint FCE training after the EBM-only initialization stage (ablation)Conv-small EBM on SVHN · SVHN, 1000 labelsclassification error (%): 8.50 -> 6.35Improved
add joint FCE training after the EBM-only initialization stage (ablation)Conv-large EBM on SVHN · SVHN, 500 labelsclassification error (%): 8.86 -> 6.86Improved
add joint FCE training after the EBM-only initialization stage (ablation)Conv-large EBM on SVHN · SVHN, 1000 labelsclassification error (%): 7.60 -> 5.54Improved
add VAT regularization (ablation)FCE with Conv-large EBM · SVHN, 500 labelsclassification error (%): 6.86 -> 4.47Improved
add VAT regularization (ablation)FCE with Conv-large EBM · SVHN, 1000 labelsclassification error (%): 5.54 -> 3.87Improved

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