Research map / Graph learning / Graph self-supervised learning
Neighborhood Reconstructing Autoencoders
Yonghyeon Lee, Hyeokjun Kwon, Frank C. Park
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 |
|---|---|---|---|
| add NRAE-L loss instead of plain AE | AE · sin(x) manifold | MSE x100: 1.90 ± 0.23 -> 0.29 ± 0.06 | Improved |
| use NRAE-L instead of DAE for denoising | DAE · CIFAR10 noise 0.1 | PSNR: 20.42 -> 22.21 | Improved |
| use NRAE-Q instead of AE on small CIFAR10 | AE · CIFAR10 S | test MSE: 0.01466 -> 0.00691 | Improved |
| use NRAE-L instead of AE on small MNIST | AE · MNIST S | test MSE: 0.01002 -> 0.00953 | Improved |
| use NRAE-L instead of CAE on small CIFAR100 | CAE · CIFAR100 S | test MSE: 0.01369 -> 0.00765 | Improved |
| combine NRAE-L loss with VAE | VAE · MNIST S | test MSE: 0.01091 -> 0.01089 | No clear effect |
| combine NRAE-L loss with VAE | VAE · CIFAR10 S | test MSE: 0.01620 -> 0.01613 | No clear effect |
| combine NRAE-L loss with WAE | WAE · CIFAR10 S | test MSE: 0.01431 -> 0.00707 | Improved |
| combine NRAE-L loss with CAE | CAE · CIFAR10 S | test MSE: 0.01208 -> 0.00718 | Improved |
| combine NRAE-L loss with SPAE | SPAE · CIFAR10 S | test MSE: 0.01504 -> 0.00723 | Improved |
| combine NRAE-L loss with GRAE | GRAE · CIFAR10 S | test MSE: 0.01452 -> 0.00716 | Improved |
| use extended NRAE-L (composition) instead of NRAE-L (ablation) | NRAE-L · MNIST S | test MSE: 0.00953 -> 0.01000 | Hurt |
Closest papers by meaning
- Graph Geometry-Preserving Autoencoders (ICML 2024 · similarity 0.88)
- Geometrically regularized autoencoders for non-Euclidean data (ICLR 2023 · similarity 0.87)
- Autoencoder Image Interpolation by Shaping the Latent Space (ICML 2021 · similarity 0.86)
- Sparse Autoencoders, Again? (ICML 2025 · similarity 0.86)
- Autoencoding Random Forests (NeurIPS 2025 · similarity 0.85)
- Bringing NeRFs to the Latent Space: Inverse Graphics Autoencoder (ICLR 2025 · similarity 0.84)
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.
