Research map / Probabilistic and scientific modeling / Variational and generative sampling
Capturing Label Characteristics in VAEs
Read the paperCheck 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 |
|---|---|---|---|
| Represent each label with continuous characteristic latents and a conditional prior, rather than making labels direct latent variables | ReVAE · CelebA and Chexpert | swap log-probability difference: CelebA f=0.004: 1.177 vs 1.489; Chexpert f=1.0: 1.084 vs 1.259 | Helped |
| Represent each label with continuous characteristic latents and a conditional prior, rather than making labels direct latent variables | ReVAE · CelebA and Chexpert | classification accuracy: CelebA f=0.06: 0.862 vs 0.862; Chexpert f=1.0: 0.826 vs 0.774 | Helped |
| Use ReVAE for conditional generation instead of M2 | ReVAE · CelebA and Chexpert | generation accuracy: CelebA f=0.004: 0.513 vs 0.499; CelebA f=1.0: 0.596 vs 0.611 | Mixed |
| Increase the classifier strength (ablation) | ReVAE · CelebA and Chexpert | classification accuracy | No clear effect |
| Do not reparameterize through characteristic latents for the label classifier (ablation) | ReVAE · training | classifier gradient-norm variance | Helped |
Closest papers by meaning
- Learning Latent Subspaces in Variational Autoencoders (NeurIPS 2018 · similarity 0.90)
- Robust Weak Supervision with Variational Auto-Encoders (ICML 2023 · similarity 0.87)
- On the Limitations of Multimodal VAEs (ICLR 2022 · similarity 0.87)
- Information-Theoretic Generalization Bounds for VAEs: A Role of Encoder and Latent Variable (ICML 2026 · similarity 0.86)
- A Statistical Analysis of Wasserstein Autoencoders for Intrinsically Low-dimensional Data (ICLR 2024 · similarity 0.86)
- Beyond Vanilla Variational Autoencoders: Detecting Posterior Collapse in Conditional and Hierarchical Variational Autoencoders (ICLR 2024 · similarity 0.85)
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.