Methods › General › Variational Optimization › Residual Normal Distribution
Residual Normal Distribution
Introduced by Arash Vahdat et al. in NVAE: A Deep Hierarchical Variational Autoencoder
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
Residual Normal Distributions are used to help the optimization of VAEs, preventing optimization from entering an unstable region. This can happen due to sharp gradients caused in situations where the encoder and decoder produce distributions far away from each other. The residual distribution parameterizes q(𝐳|𝐱) relative to p(𝐳). Let p(zⁱₗ|𝐳_(<l)) := N (μᵢ(𝐳_(<l)), σᵢ(𝐳_(<l))) be a Normal distribution for the $i$th variable in 𝐳ₗ in prior. Define q(zⁱₗ|𝐳_(<l), x) := N(μᵢ(𝐳_(<l)) + Δμᵢ(𝐳_(<l), x), σᵢ(𝐳_(<l)) ·Δσᵢ(𝐳_(<l), x) ), where Δμᵢ(𝐳_(<l), 𝐱) and Δσᵢ(𝐳_(<l), 𝐱) are the relative location and scale of the approximate posterior with respect to the prior. With this parameterization, when the prior moves, the approximate posterior moves accordingly, if not changed.
Papers archive 2025-07-28
5 shown of 5, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.
-
A Variational AutoEncoder for Transformers with Nonparametric Variational Information Bottleneck 27 Jul 2022 · 0 repositories · arXiv:2207.13529
-
Alleviating Adversarial Attacks on Variational Autoencoders with MCMC 18 Mar 2022 · 1 repository · arXiv:2203.09940
-
Polarity Sampling: Quality and Diversity Control of Pre-Trained Generative Networks via Singular Values 3 Mar 2022 · 1 repository · arXiv:2203.01993
-
NVAE-GAN Based Approach for Unsupervised Time Series Anomaly Detection 8 Jan 2021 · 0 repositories · arXiv:2101.02908
-
NVAE: A Deep Hierarchical Variational Autoencoder 8 Jul 2020 · 10 repositories · arXiv:2007.03898Syntology ran 23 of 41 samples · 18 unverified · 23 pointer-only (licence)
Tasks archive 2025-07-28
9 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
| Task | Papers |
|---|---|
| Image Generation | 2 |
| Adversarial Attack | 1 |
| Anomaly Detection | 1 |
| Decoder | 1 |
| Diversity | 1 |
| Time Series | 1 |
| Time Series Analysis | 1 |
| Time Series Anomaly Detection | 1 |
| Unconditional Image Generation | 1 |
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections