Research map / Learning theory and algorithms / Optimal transport
Wasserstein Iterative Networks for Barycenter Estimation
Alexander Korotin, Vage Egiazarian, Lingxiao Li, Evgeny Burnaev
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| replace ICNN-based SCW2B with WIN (generator output) | barycenter estimation · Ave, celeba! | FID: 156.3 -> 75.8 | Improved |
| replace SCW2B with WIN, weighted mapped output | barycenter estimation · Ave, celeba! | FID: 152.1 -> 52.85 | Improved |
| replace SCW2B with WIN for inverse maps, n=3 | inverse OT maps · Ave, celeba! | FID: 319.62 -> 61.5 | Improved |
| replace SCW2B with WIN for inverse maps, n=1 | inverse OT maps · Ave, celeba! | FID: 67.4 -> 49.3 | Improved |
| replace constant-shift baseline with WIN inverse maps, n=2 | inverse OT maps · Ave, celeba! | FID: 75.5 -> 46.9 | Improved |
| replace SCW2B with WIN | barycenter, Gaussian location-scatter D=128 · BW2-UVP | BW2-UVP (%): 1.28 -> 0.38 | Improved |
| replace SCW2B with WIN | barycenter, Uniform location-scatter D=128 · BW2-UVP | BW2-UVP (%): 1.38 -> 0.46 | Improved |
| replace SCW2B with WIN | barycenter, Gaussian location-scatter D=2 · BW2-UVP | BW2-UVP (%): 0.07 -> 0.01 | Improved |
| replace ICNN-based SCW2B with WIN as N=1 generative model | CelebA generation · CelebA 64x64 | FID: 90.2 -> 46.6 | Improved |
| compare WIN to WGAN-QC as N=1 generative model (mapped output) | CelebA generation · CelebA 64x64 | FID: 14.4 vs 15.7 | No clear effect |
| fix randomly initialized generator, train only the map (ablation) | WIN, N=1 · CelebA 64x64 | FID: 15.7 -> 16.1 | No clear effect |
| map generated samples through learned input maps (weighted sum) | WIN · Ave, celeba! | FID: 75.8 -> 52.85 | Improved |
Closest papers by meaning
- Continuous Wasserstein-2 Barycenter Estimation without Minimax Optimization (ICLR 2021 · similarity 0.93)
- Scalable Computations of Wasserstein Barycenter via Input Convex Neural Networks (ICML 2021 · similarity 0.93)
- Estimating Barycenters of Distributions with Neural Optimal Transport (ICML 2024 · similarity 0.89)
- A Unified Approach for Computing Wasserstein Barycenters of Discrete and Continuous Measures (NeurIPS 2026 · similarity 0.88)
- Finding Wasserstein Ball Center: Efficient Algorithm and The Applications in Fairness (ICML 2025 · similarity 0.87)
- Optimal Transport Barycenter via Nonconvex-Concave Minimax Optimization (ICML 2025 · 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.
