AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Exploring Bidirectional Bounds for Minimax-Training of Energy-Based Models

Cong Geng, Jia Wang, Li Chen, Zhiyong Gao, Jes Frellsen, Søren Hauberg

IJCV 2025 · Learning theory and algorithms › Kernels and regression theory · approach: Information-theoretic generalization bounds · 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
replace zero-centered gradient penalty with the derived GP upper bound (ablation)EBM SV lower bound · ANIMEFACEFID: 20.53 -> 12.75Improved
replace zero-centered gradient penalty with the derived GP upper bound (ablation)EBM SV lower bound · CIFAR-10FID: 35.42 -> 28.63Improved
use a true upper bound instead of zero-centered GP (ablation)EBM generator · ANIMEFACEanisotropy index: 3.4560 -> 1.8016Improved
use GP upper bound instead of zero-centered GP with SV lower bound (ablation)EBM on normalizing flow · MNISTtest log-likelihood: -2167 -> -1634Improved
use GP upper bound instead of zero-centered GP with MI lower bound (ablation)EBM on normalizing flow · MNISTtest log-likelihood: -1819 -> -1782Improved
use diffusion upper bound with SV lower bound instead of GP (ablation)EBM on normalizing flow · MNISTtest log-likelihood: -1634 -> -1561Improved
swap MI lower bound for SV lower bound (GP upper) (ablation)EBM · CIFAR-10FID: 28.96 -> 28.63No clear effect
raise entropy weight lambda from 0.0001 to 1 (ablation)EBM SV+GP · CIFAR-10FID: 28.63 -> 45.08Hurt
raise entropy weight lambda from 0.0001 to 1 (ablation)EBM SV+Diff · CIFAR-10FID: 29.31 -> 33.13Hurt
scale network from DCGAN to UNetEBM MI+Diff · CIFAR-10FID: 30.37 -> 5.47Improved
scale to large UNetEBM MI+Diff · ImageNet 32x32FID: 14.61 -> 6.57Improved
add bidirectional bounds (SV+GP) versus VERA lower-bound-only minimaxEBM · CIFAR-10FID: 66.38 -> 28.63Improved

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