Research map / Probabilistic and scientific modeling / Variational and generative sampling
Structured latent VAEs
Improves VAEs with dependent, discrete and Laplace-based latent structure and generalization bounds.
125 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Variational and generative sampling, keyword cluster “vae · autoencoders · latent”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
22 papers in 2023–24 and 22 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 5.3% to 4.4%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Auxiliary Deep Generative Models (ICML 2016 · 257 citations)
- 2017 · Neural Discrete Representation Learning (NeurIPS 2017 · 2,816 citations)
- 2018 · A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music (ICML 2018 · 301 citations)
- 2019 · InfoVAE: Balancing Learning and Inference in Variational Autoencoders (AAAI 2019 · 274 citations)
- 2020 · Improving Inference for Neural Image Compression (NeurIPS 2020 · 54 citations)
- 2021 · Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech (ICML 2021 · 121 citations)
- 2022 · SQ-VAE: Variational Bayes on Discrete Representation with Self-annealed Stochastic Quantization (ICML 2022 · 13 citations)
- 2023 · Understanding Masked Autoencoders via Hierarchical Latent Variable Models (CVPR 2023 · 30 citations)
- 2024 · Unity by Diversity: Improved Representation Learning for Multimodal VAEs (NeurIPS 2024 · 4 citations)
- 2025 · Latent Thought Models with Variational Bayes Inference-Time Computation (ICML 2025 · 1 citation)
- 2026 · Training-Free Vector Quantization via Gaussian VAEs (ICML 2026 · 0 citations)
Earliest papers here
- Auxiliary Deep Generative Models (ICML 2016 · 257 citations)
- Ladder Variational Autoencoders (NeurIPS 2016 · 246 citations)
- Neural Discrete Representation Learning (NeurIPS 2017 · 2,816 citations)
What has been tried
97 papers here report 939 measured changes (a component added, removed or swapped, compared with a baseline): helped 66%, no clear effect 7%, hurt 22%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Use one thought-vector set across all layers instead of distinct layered sets | language modeling · reconstruction accuracy | Hurt | Latent Thought Models with Variational Bayes Inference-Time Computation (ICML 2025 · 1 citation) | |
| replace the causal posterior encoder with a non-causal posterior encoder | Opencpop ablation · F0 RMSE | 28.101 -> 28.334 | Hurt | CSSinger: End-to-End Chunkwise Streaming Singing Voice Synthesis System Based on Conditional Variational Autoencoder (AAAI 2025 · 0 citations) |
| increase latent dimensions from 2 to 10 | TSPLIB test sets · edge recall | 0.950 ± 0.007 -> 0.985 ± 0.002; 0.883 ± 0.013 -> 0.971 ± 0.005; 0.810 ± 0.009 -> 0.960 ± 0.004; 0.723 ± 0. | Helped | Inverse Optimization Latent Variable Models for Learning Costs Applied to Route Problems (NeurIPS 2025 · 0 citations) |
| add the FVAE factorization regularizer | Image classification average (DTD, EuroSAT, GTSRB, RESISC45, SUN397, SVHN) · top-1 accuracy | 86.43 -> 89.53 | Helped | Latent Space Factorization in LoRA (NeurIPS 2025 · 0 citations) |
| Use ST Gumbel monotonic alignment instead of cross-attention | LibriSpeech test, alignment ablation · WER (%) | 6.6 -> 2.72 | Helped | Continuous Autoregressive Modeling with Stochastic Monotonic Alignment for Speech Synthesis (ICLR 2025 · 0 citations) |
| Increase ν from moderate values toward very small values | Low-dimensional synthetic data · generation quality | Hurt | $t^3$-Variational Autoencoder: Learning Heavy-tailed Data with Student's t and Power Divergence (ICLR 2024 · 0 citations) | |
| Increase the importance-sampling particle count K | DS1–DS4 · KL divergence | DS1, d=2: 0.1275 -> 0.0273 (K=1 to K=32); DS2, d=2: 0.0202 -> 0.0100; DS3, d=10: 0.0980 -> 0.0477; DS4, d=2: 0.1038 -> 0 | Helped | PhyloVAE: Unsupervised Learning of Phylogenetic Trees via Variational Autoencoders (ICLR 2025) |
What reviewers will expect
Datasets and baselines named in the 13 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- MNIST · 2 of 13 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 44 papers; from OpenAlex and the papers' affiliations).
- ETHZ - ETH Zurich · 3 papers
- Swiss Federal Institute of Technology · 3 papers
- Idiap Research Institute · 2 papers
- Seoul National University · 2 papers
- RIKEN · 2 papers
- Google DeepMind · 2 papers
- University of Oxford · 2 papers
- University of Pennsylvania · 2 papers
- University of California, Los Angeles · 2 papers
Use it
Download BibTeX of 40 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Regularizing VAE latent information (89 papers)
- Disentangled and discrete latent models (92 papers)
- Hierarchical VAEs and posterior collapse (59 papers)
Elsewhere in the topic
- Likelihood-free and graphical model estimation (161 papers)
- Flow matching and flow sampling (144 papers)
- Posterior sampling with diffusion guidance (137 papers)
- Score-based generative modeling analysis (128 papers)
- Exact inference for probabilistic programs (103 papers)
- Latent variable marginalization and estimators (91 papers)
- Tractable probabilistic models (90 papers)
- Variational inference theory and gradients (89 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).