Research map / Image generation / GANs
Topology Distance: A Topology-Based Approach for Evaluating Generative Adversarial Networks
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 Topology Distance instead of FID or KID to compare a Gaussian with a two-Gaussian mixture having the same first and second moments | GAN evaluation metrics · Synthetic Gaussian distributions | distribution discrimination | Helped |
| Use Topology Distance instead of FID or KID to evaluate increasingly corrupted images | GAN evaluation metrics · CelebA manipulated images | consistency with image-quality degradation | Helped |
| Use Topology Distance instead of Geometry Score to measure perturbation severity | GAN evaluation metrics · CIFAR100-C | perturbation-severity consistency | Helped |
| Use Topology Distance instead of Inception Score to track generated-image quality | GAN evaluation metrics · CelebA | correlation with generated-image quality | Helped |
| Use learned features instead of raw pixels to compute Topology Distance (ablation) | Topology Distance · Fashion-MNIST and CIFAR10 | discrimination and consistency | Helped |
| Use Topology Distance instead of Geometry Score to distinguish a Gaussian from a two-Gaussian mixture | GAN evaluation metrics · Synthetic Gaussian distributions | distribution discrimination | Helped |
Closest papers by meaning
- Geometry Score: A Method For Comparing Generative Adversarial Networks (ICML 2018 · similarity 0.92)
- A Geometric Analysis of Deep Generative Image Models and Its Applications (ICLR 2021 · similarity 0.90)
- Manifold Topology Divergence: a Framework for Comparing Data Manifolds. (NeurIPS 2021 · similarity 0.89)
- Unveiling the Latent Space Geometry of Push-Forward Generative Models (ICML 2023 · similarity 0.88)
- Quantitatively Evaluating GANs With Divergences Proposed for Training (ICLR 2018 · similarity 0.88)
- SMaRt: Improving GANs with Score Matching Regularity (ICML 2024 · similarity 0.87)
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.