Research map / Probabilistic and scientific modeling / Variational and generative sampling
Wasserstein Variational Inference
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 Wasserstein variational autoencoder 1111 instead of VAE | MNIST · MNIST (30 reruns) | Observable mean squared reconstruction error: 0.042 ± 0.003 -> 0.029 ± 0.007 | Helped |
| Use Wasserstein variational autoencoder 1111 instead of ALI | MNIST · MNIST (30 reruns) | Sample quality distance: 0.057 ± 0.002 -> 0.041 ± 0.002 | Helped |
| Use 1111 instead of VAE | Wasserstein variational autoencoder · Fashion-MNIST | Observable: 0.0214 -> 0.0258 | Hurt |
| Use 1111 instead of VAE | Wasserstein variational autoencoder · Quick Sketch | Sample: 0.0687 -> 0.0699 | Mixed |
| Use only cost components one and four (1001) instead of all four (ablation) | Wasserstein variational autoencoder 1111 · MNIST | Observable: 0.0293 -> 0.0710 | Hurt |
| Use only cost components two and three (0110) instead of all four (ablation) | Wasserstein variational autoencoder 1111 · MNIST | Observable: 0.0293 -> 0.0227 | Helped |
| Use only cost components three and four (0011) instead of all four (ablation) | Wasserstein variational autoencoder 1111 · MNIST | Observable: 0.0293 -> 0.0273 | Helped |
| Use only cost components one and two (1100) instead of all four (ablation) | Wasserstein variational autoencoder 1111 · MNIST | Observable: 0.0293 -> 0.0268 | Helped |
| Omit explicit marginal training of x (0011) (ablation) | Wasserstein variational autoencoder · MNIST, Fashion-MNIST, Quick Sketch | Sample quality: 0.0441 -> 0.0740; 0.0297 -> 0.0447; 0.0699 -> 0.1204 | Hurt |
| Add Sinkhorn loss to the VAE or ALI objective | Hybrid autoencoders · MNIST, Fashion-MNIST, Quick Sketch | Performance across reported metrics | Mixed |
Closest papers by meaning
- Generalized Variational Inference via Optimal Transport (AAAI 2024 · similarity 0.92)
- Stochastic variance-reduced Gaussian variational inference on the Bures-Wasserstein manifold (ICLR 2025 · similarity 0.88)
- Asymptotics of Alpha-Divergence Variational Inference Algorithms with Exponential Families (NeurIPS 2024 · similarity 0.88)
- Bayesian Online Natural Gradient (BONG) (NeurIPS 2024 · similarity 0.87)
- Maxitive Donsker-Varadhan Formulation for Possibilistic Variational Inference (NeurIPS 2026 · similarity 0.87)
- Transport meets Variational Inference: Controlled Monte Carlo Diffusions (ICLR 2024 · similarity 0.86)
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.