Research map / Probabilistic and scientific modeling / Variational and generative sampling
Trading Information between Latents in Hierarchical Variational Autoencoders
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 layer-wise rate trade-offs (beta2 != beta1) instead of a conventional beta-VAE (beta2=beta1) | GHVAE on SVHN · PSNR reconstruction, constant total rate | PSNR at fixed total rate | Helped |
| Shift rate toward z1 (downstream layer) rather than z2 for classification with less expressive classifiers or complex data | GHVAE on CIFAR-10 representation learning · linear classifiers / CIFAR-10 mu_2 | classification accuracy | Helped |
| Increase rate R(z2) (higher mutual information) for expressive classifiers on simple data | GHVAE on SVHN representation learning · RBF SVM / kNN on mu_2 | classification accuracy | Helped |
| Increase rate in z2 with beta2 < beta1 (high R(z2), low R(z1|z2)) for very high rates | GHVAE on SVHN sample generation · Inception Score | Inception Score | Mixed |
| Higher total rate (lower beta) for reconstruction | GHVAE on SVHN and CIFAR-10 · PSNR | PSNR | Helped |
| Use the same beta for all layers (conventional beta-VAE) vs. separately tuned layer betas for generative modeling on MNIST | GHVAE on MNIST · Inception Score | Inception Score | No clear effect |
Closest papers by meaning
- Hierarchical Decompositional Mixtures of Variational Autoencoders (ICML 2019 · similarity 0.93)
- Learning Hierarchical Priors in VAEs (NeurIPS 2019 · similarity 0.92)
- Variational Autoencoders with Jointly Optimized Latent Dependency Structure (ICLR 2019 · similarity 0.92)
- Information-Theoretic Generalization Bounds for VAEs: A Role of Encoder and Latent Variable (ICML 2026 · similarity 0.92)
- From Variational to Deterministic Autoencoders (ICLR 2020 · similarity 0.92)
- Beyond Vanilla Variational Autoencoders: Detecting Posterior Collapse in Conditional and Hierarchical Variational Autoencoders (ICLR 2024 · similarity 0.91)
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.