Research map / Probabilistic and scientific modeling / Variational and generative sampling
Multi-Level Variational Autoencoder: Learning Disentangled Representations From Grouped Observations
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 |
|---|---|---|---|
| Use grouped observations to infer shared content and per-observation style | ML-VAE · MNIST | digit-class information in content and style | Helped |
| Use grouped observations to infer shared content and per-observation style | ML-VAE · MS-Celeb-1M test identities | identity information in content and style | Helped |
| Accumulate multiple test images from the same group when constructing content (ablation) | ML-VAE · MS-Celeb-1M reconstruction | reconstruction quality | Helped |
| Accumulate more images per class to construct the content representation (ablation) | ML-VAE · MNIST and MS-Celeb-1M classification | content-feature classification performance | Helped |
| Accumulate more images per class to construct the content representation (ablation) | ML-VAE · MNIST and MS-Celeb-1M classification | style-feature classification performance | No clear effect |
| Replace the original VAE representation with ML-VAE content features | VAE · MNIST and MS-Celeb-1M classification | content-feature classification accuracy and conditional entropy | No clear effect |
Closest papers by meaning
- Disentangling Disentanglement in Variational Autoencoders (ICML 2019 · similarity 0.92)
- Multi-Facet Clustering Variational Autoencoders (NeurIPS 2021 · similarity 0.90)
- Learning Disentangled Representations with Semi-Supervised Deep Generative Models (NeurIPS 2017 · similarity 0.90)
- Towards Building A Group-based Unsupervised Representation Disentanglement Framework (ICLR 2022 · similarity 0.90)
- An Identifiable Double VAE For Disentangled Representations (ICML 2021 · similarity 0.89)
- $\alpha$TC-VAE: On the relationship between Disentanglement and Diversity (ICLR 2024 · similarity 0.88)
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.