Research map / Probabilistic and scientific modeling / Variational and generative sampling
Towards Visually Explaining Variational Autoencoders
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 |
|---|---|---|---|
| Use Conv2 rather than reconstruction difference for anomaly localization | VAE attention on UCSD Ped1 · UCSD Ped1 | pixel-level AUROC: 0.86 -> 0.92 | Helped |
| Target Conv1 features for attention instead of Conv2 (ablation) | VAE anomaly attention · UCSD Ped1 | pixel-level AUROC: 0.92 -> 0.89 | Hurt |
| Target Conv3 features for attention instead of Conv2 (ablation) | VAE anomaly attention · UCSD Ped1 | pixel-level AUROC: 0.92 -> 0.91 | Hurt |
| Add attention disentanglement loss | FactorVAE · dSprites | disentanglement score: around 0.82 -> around 0.90 | Helped |
| Use AD-FactorVAE instead of β-VAE (β=4) | dSprites disentanglement · dSprites | disentanglement score: 0.73 -> around 0.90 | Helped |
| Use VAE attention localization instead of AE (SSIM) | MVTec-AD anomaly localization · MVTec-AD | pixel-level ROC AUC | Mixed |
Closest papers by meaning
- Explaining Deep Convolutional Neural Networks via Latent Visual-Semantic Filter Attention (CVPR 2022 · similarity 0.89)
- Spatially Informed Autoencoders for Interpretable Visual Representation Learning (ICLR 2026 · similarity 0.88)
- beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework (ICLR 2017 · similarity 0.88)
- Latent Space Explanation by Intervention (AAAI 2022 · similarity 0.87)
- On Guiding Visual Attention With Language Specification (CVPR 2022 · similarity 0.87)
- Attention Guided CAM: Visual Explanations of Vision Transformer Guided by Self-Attention (AAAI 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.