Research map / Probabilistic and scientific modeling / Variational and generative sampling
A Forest from the Trees: Generation through Neighborhoods
Yang Li, Tianxiang Gao, Junier Oliva
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 |
|---|---|---|---|
| Use NCT instead of RealNVP | RealNVP · MNIST | FID: 10.4 -> 8.2 | Improved |
| Use NCL instead of RealNVP | RealNVP · SVHN | FID: 103.8 -> 50.0 | Improved |
| Use NCT instead of RealNVP | RealNVP · CIFAR-10 | FID: 99.9 -> 81.1 | Improved |
| Use NCT instead of RealNVP | RealNVP · CelebA | FID: 39.2 -> 30.9 | Improved |
| Use NCT instead of RealNVP | RealNVP · CIFAR-10 | PRD (F8, F1/8): (0.322, 0.631) -> (0.568, 0.691) | Improved |
| Use NCL instead of RealNVP | RealNVP · SVHN | PRD (F8, F1/8): (0.740, 0.731) -> (0.808, 0.928) | Improved |
| Use NCL instead of RealNVP | RealNVP · MNIST | PRD (F8, F1/8): (0.982, 0.982) -> (0.981, 0.970) | Mixed |
| Use neighborhood-conditioned flow | RealNVP · CelebA | PRD (F8, F1/8): RNVP (0.675, 0.766); NCL (0.853, 0.806); NCT (0.849, 0.821) | Mixed |
| Add contrastive training (ablation) | NCL · MNIST novelty detection (mean over classes) | AUC-ROC: 0.621 -> 0.801 | Improved |
| Remove contrastive training (ablation) | NCL · MNIST novelty detection (mean over classes) | AUC-ROC: 0.801 -> 0.621 | Hurt |
| Use NCL instead of RealNVP | RealNVP · MNIST novelty detection (mean over classes) | AUC-ROC: 0.641 -> 0.621 | Hurt |
| Use NCL with contrastive training instead of RealNVP | RealNVP · MNIST novelty detection (mean over classes) | AUC-ROC: 0.641 -> 0.801 | Improved |
Closest papers by meaning
- Generative Models as a Data Source for Multiview Representation Learning (ICLR 2022 · similarity 0.87)
- Is Generation Required for Data-Efficient Perception? (ICML 2026 · similarity 0.86)
- Analysis of Learning a Flow-based Generative Model from Limited Sample Complexity (ICLR 2024 · similarity 0.86)
- Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective (NeurIPS 2024 · similarity 0.86)
- Linear combinations of latents in generative models: subspaces and beyond (ICLR 2025 · similarity 0.85)
- Boosting Generative Image Modeling via Joint Image-Feature Synthesis (NeurIPS 2025 · similarity 0.85)
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.
