Research map / Label-efficient and robust learning / Domain adaptation and generalization
Multivariate-Information Adversarial Ensemble for Scalable Joint Distribution Matching
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 MMI-ALI instead of CycleGAN | Rotated MNIST supervised generation · Rotated MNIST, 1% supervision | SSIM: 0.32 -> 0.57 | Helped |
| Use MMI-ALI instead of CycleGAN | Rotated MNIST supervised generation · Rotated MNIST, 5% supervision | SSIM: 0.31 -> 0.68 | Helped |
| Use MMI-ALI instead of CycleGAN | Rotated MNIST supervised generation · Rotated MNIST, 10% supervision | SSIM: 0.35 -> 0.72 | Helped |
| Use MMI-ALI instead of CycleGAN | Rotated MNIST unsupervised generation · Rotated MNIST, six transfer directions | Inception Score: 8.34 -> 8.99; 6.13 -> 9.01; 2.25 -> 2.95; 2.38 -> 3.86; 1.71 -> 3.31; 1.04 -> 3.08 | Helped |
| Use MMI-ALI instead of CycleGAN | Cityscape unsupervised transfer · Cityscape, six directions | FID: 224.04 -> 202.93; 213.43 -> 254.41; 164.65 -> 150.98; 222.24 -> 246.04; 60.20 -> 101.30; 144.07 -> 192.13 | Mixed |
| Use MMI-ALI instead of CycleGAN | MojiTalk emotion transfer · MojiTalk, Happy | MRR: 0.71 -> 1.0 | Helped |
| Use MMI-ALI instead of the groundtruth reference | MojiTalk emotion transfer · MojiTalk, Angry | MRR: 0.41 -> 0.40 | Hurt |
| Use MMI-ALI instead of CycleGAN | Three-domain object transfiguration · Zebra, Horse, and Okapi transfers | human preference: 52.1%–67.2% | Helped |
| Add limited paired supervision | MMI-ALI · Rotated MNIST supervised transfer | SSIM: 0.57; 0.68; 0.72 | Helped |
Closest papers by meaning
- JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets (ICML 2018 · similarity 0.89)
- ALICE: Towards Understanding Adversarial Learning for Joint Distribution Matching (NeurIPS 2017 · similarity 0.87)
- Iterative Feature Matching: Toward Provable Domain Generalization with Logarithmic Environments (NeurIPS 2022 · similarity 0.86)
- Triangle Generative Adversarial Networks (NeurIPS 2017 · similarity 0.85)
- Expressive Score-Based Priors for Distribution Matching with Geometry-Preserving Regularization (ICML 2025 · similarity 0.83)
- Efficient Distribution Matching of Representations via Noise-Injected Deep InfoMax (ICLR 2025 · similarity 0.83)
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.