AI Paper Atlas

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

Energy Discrepancies: A Score-Independent Loss for Energy-Based Models

Tobias Schröder, Zijing Ou, Jen Ning Lim, Yingzhen Li, Sebastian Josef Vollmer, Andrew Duncan

NeurIPS 2023 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Posterior sampling with diffusion guidance · 0 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
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · SVHNFID: 29.44 -> 28.10Improved
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · CelebA 64x64FID: 37.87 -> 36.73Improved
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · CIFAR-10FID: 70.15 -> 73.58Hurt
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · CIFAR-10MSE: 0.020 -> 0.023Hurt
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · SVHNMSE: 0.008 -> 0.006Improved
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · CelebA 64x64MSE: 0.013 -> 0.009Improved
train latent EBM with energy discrepancy instead of score matchingED-LEBM · SVHNFID: 34.44 -> 28.10Improved
train latent EBM with energy discrepancy instead of score matchingED-LEBM · CIFAR-10FID: 77.82 -> 73.58Improved
train latent EBM with energy discrepancy instead of score matchingED-LEBM · CelebA 64x64FID: 41.21 -> 36.73Improved
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · MNIST anomaly, heldout digit 4AUPRC: 0.630 -> 0.740Improved
train latent EBM with energy discrepancy instead of contrastive divergenceED-LEBM · MNIST anomaly, heldout digit 1AUPRC: 0.336 -> 0.342Improved
train latent EBM with energy discrepancy instead of score matchingED-LEBM · MNIST anomaly, heldout digit 1AUPRC: 0.285 -> 0.342Improved

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