Research map / Probabilistic and scientific modeling / Variational and generative sampling
Unveiling the Latent Space Geometry of Push-Forward Generative Models
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 |
|---|---|---|---|
| simplicial cluster latent truncation (rejection sampling on latent) | TransGAN · CIFAR-10 | Precision: +2.1 (72.8 to 74.9) | Helped |
| simplicial cluster latent truncation (rejection sampling on latent) | TransGAN · CIFAR-10 | Density: +17.1 (79.3 to 96.4) | Helped |
| simplicial cluster latent truncation (rejection sampling on latent) | TransGAN · CIFAR-10 | FID: +0.3 (8.9 to 9.2) | Hurt |
| simplicial cluster latent truncation (rejection sampling on latent) | TransGAN · CIFAR-100 | Precision: +1.4 (64.2 to 65.6) | Helped |
| simplicial cluster latent truncation (rejection sampling on latent) | TransGAN · STL-10 (32x32) | Precision: +2.1 (75.7 to 77.8) | Helped |
| simplicial cluster latent truncation (rejection sampling on latent) | TransGAN · STL-10 (32x32) | Density: +6.6 (87.5 to 94.1) | Helped |
| simplicial cluster latent truncation vs. JBT and DeliGAN | TransGAN · CIFAR-10 | Precision: simp. 74.9 vs DeliGAN 74.6 and JBT 73.3 | Mixed |
| latent-space linear separability and convexity correlate with precision (width-scaling study) (ablation) | GAN latent space on CIFAR-10 · CIFAR-10 | Pearson correlation Precision vs LogReg Accuracy: r = 0.98 | Helped |
| increasing latent dimension beyond the number of modes (ablation) | GANs · Multiple datasets (latent dim 2 to 128) | Precision | Helped |
Closest papers by meaning
- Complexity Matters: Rethinking the Latent Space for Generative Modeling (NeurIPS 2023 · similarity 0.92)
- Latent Space Oddity: on the Curvature of Deep Generative Models (ICLR 2018 · similarity 0.92)
- Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective (NeurIPS 2024 · similarity 0.87)
- Linear combinations of latents in generative models: subspaces and beyond (ICLR 2025 · similarity 0.87)
- GLAD: Improving Latent Graph Generative Modeling with Simple Quantization (AAAI 2025 · similarity 0.87)
- What Secrets Do Your Manifolds Hold? Understanding the Local Geometry of Generative Models (ICLR 2025 · 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.