Research map / Probabilistic and scientific modeling / Variational and generative sampling
An Identifiable Double VAE For Disentangled Representations
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 an optimal learned conditional prior instead of the iVAE conditional prior | IDVAE vs iVAE · dsprites, cars3d, shapes3d, smallnorb | disentanglement scores | Helped |
| Use IDVAE instead of the competing methods | disentanglement methods · dsprites, cars3d, shapes3d | beta score | Helped |
| Use IDVAE instead of FullVAE | disentanglement methods · smallnorb | beta score | Hurt |
| Provide more labeled instances | SS-IDVAE · dsprites, cars3d, shapes3d | disentanglement performance | Helped |
| Provide more labeled instances | SS-iVAE · disentanglement datasets | disentanglement performance | Helped |
| Provide more labeled instances | SS-FullVAE · disentanglement datasets | disentanglement performance | No clear effect |
| Train with 1% labeled samples instead of no auxiliary information | SS-IDVAE vs β-VAE · disentanglement datasets | disentanglement performance | Helped |
| Increase regularization strength | IDVAE · dsprites, cars3d | beta score | No clear effect |
| Increase regularization strength | IDVAE · shapes3d, smallnorb | beta score | Hurt |
| Increase regularization strength | iVAE · dsprites, smallnorb | beta score | Hurt |
Closest papers by meaning
- Learning Disentangled Representations with Semi-Supervised Deep Generative Models (NeurIPS 2017 · similarity 0.93)
- Disentangling Disentanglement in Variational Autoencoders (ICML 2019 · similarity 0.92)
- Guided Variational Autoencoder for Disentanglement Learning (CVPR 2020 · similarity 0.92)
- Demystifying Inductive Biases for (Beta-)VAE Based Architectures (ICML 2021 · similarity 0.92)
- $\alpha$TC-VAE: On the relationship between Disentanglement and Diversity (ICLR 2024 · similarity 0.90)
- Why do Variational Autoencoders Really Promote Disentanglement? (ICML 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.