Research map / Probabilistic and scientific modeling / Variational and generative sampling
MAE: Mutual Posterior-Divergence Regularization for 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 |
|---|---|---|---|
| Add mutual posterior-divergence regularization to VLAE | VLAE · MNIST dynamically binarized | NLL: 78.26 -> 77.98 | Helped |
| Add mutual posterior-divergence regularization to VLAE | VLAE · OMNIGLOT dynamically binarized | NLL: 89.62 -> 89.09 | Helped |
| Add mutual posterior-divergence regularization to VLAE | VLAE · CIFAR-10 | bits/dim: 2.98 -> 2.95 | No clear effect |
| Add mutual posterior-divergence regularization to VLAE | VLAE · MNIST K-Means K=10 | clustering accuracy: 68.1 -> 91.2 | Helped |
| Add mutual posterior-divergence regularization to VLAE | VLAE · MNIST KNN 100 labels | accuracy: 75.7 -> 86.7 | Helped |
| Add mutual posterior-divergence regularization to VLAE | VLAE · MNIST Linear 100 labels | accuracy: 86.4 -> 91.5 | Helped |
| Increase η from 0.5 to 2.0 at γ=0.5 (ablation) | MAE · MNIST K-Means K=10 | clustering accuracy: 84.7 -> 78.2 | Hurt |
| Increase η from 0.5 to 2.0 at γ=0.5 (ablation) | MAE · MNIST dynamically binarized | NLL: 78.00 -> 77.99 | Helped |
| Increase γ from 0.1 to 0.5 at η=1.0 (ablation) | MAE · MNIST dynamically binarized | NLL: 78.02 -> 77.98 | Helped |
| Increase γ from 0.5 to 1.0 at η=1.0 (ablation) | MAE · MNIST dynamically binarized | NLL: 77.98 -> 78.15 | Hurt |
| Increase γ from 0.1 to 0.5 at η=0.5 (ablation) | MAE · CIFAR-10 | bits/dim: 2.95 -> 2.97 | Hurt |
| Increase γ from 0.1 to 0.5 at η=0.5 (ablation) | MAE · OMNIGLOT dynamically binarized | NLL: 89.21 -> 89.41 | Hurt |
Closest papers by meaning
- Variational Autoencoder with Implicit Optimal Priors (AAAI 2019 · similarity 0.93)
- Consistency Regularization for Variational Auto-Encoders (NeurIPS 2021 · similarity 0.93)
- From Variational to Deterministic Autoencoders (ICLR 2020 · similarity 0.92)
- Marginalization is not Marginal: No Bad VAE Local Minima when Learning Optimal Sparse Representations (ICML 2023 · similarity 0.92)
- Beyond Vanilla Variational Autoencoders: Detecting Posterior Collapse in Conditional and Hierarchical Variational Autoencoders (ICLR 2024 · similarity 0.90)
- EQ-VAE: Equivariance Regularized Latent Space for Improved Generative Image Modeling (ICML 2025 · similarity 0.90)
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.