Research map / Probabilistic and scientific modeling / Variational and generative sampling
Joint Training of Variational Auto-Encoder and Latent Energy-Based Model
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| jointly train the VAE generator and latent EBM with the divergence-triangle objective | VAE · CIFAR-10 | FID: 109.5 -> 30.1 | Helped |
| jointly train the VAE generator and latent EBM with the divergence-triangle objective | VAE · CelebA | FID: 99.09 -> 24.7 | Helped |
| jointly train the VAE generator and latent EBM with the divergence-triangle objective | VAE · LSUN | FID: 175.2 -> 27.3 | Helped |
| use joint VAE-latent EBM training instead of VAE alone | VAE · CIFAR-10 | RMSE: 0.192 -> 0.177 | Helped |
| use joint VAE-latent EBM training instead of VAE alone | VAE · CelebA | RMSE: 0.197 -> 0.190 | Helped |
| use joint VAE-latent EBM training instead of VAE alone | VAE · LSUN-bedroom | RMSE: 0.164 -> 0.169 | No clear effect |
| use joint VAE-latent EBM training | CIFAR-10 OOD detector · SVHN OOD | AUROC: 0.63 -> 0.68 | Helped |
| use joint VAE-latent EBM training | CIFAR-10 OOD detector · Texture OOD | AUROC: 0.5 -> 0.56 | Helped |
| use joint VAE-latent EBM training | CIFAR-10 OOD detector · CelebA OOD | AUROC: 0.52 -> 0.56 | Helped |
| use joint VAE-latent EBM training | MNIST anomaly detector · MNIST holdout 4 | AUPRC: 0.443 $\pm$ 0.029 -> 0.723 $\pm$ 0.042 | Helped |
| train the inference model using the alternative objective (ablation) | Joint VAE-latent EBM · CIFAR-10 | FID: 30.1 -> 33.3 | Hurt |
| train the inference model using the alternative objective (ablation) | Joint VAE-latent EBM · CelebA | FID: 24.7 -> 29.5 | Hurt |
Closest papers by meaning
- Adversarial Symmetric Variational Autoencoder (NeurIPS 2017 · similarity 0.91)
- Learning Energy-Based Model with Variational Auto-Encoder as Amortized Sampler (AAAI 2021 · similarity 0.88)
- Learning Joint Latent Space EBM Prior Model for Multi-Layer Generator (CVPR 2023 · similarity 0.88)
- Scalable Energy-Based Models via Adversarial Training: Unifying Discrimination and Generation (ICLR 2026 · similarity 0.88)
- Improving Adversarial Energy-Based Model via Diffusion Process (ICML 2024 · similarity 0.87)
- $t^3$-Variational Autoencoder: Learning Heavy-tailed Data with Student's t and Power Divergence (ICLR 2024 · similarity 0.86)
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.