AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Neighborhood Reconstructing Autoencoders

Yonghyeon Lee, Hyeokjun Kwon, Frank C. Park

NeurIPS 2021 · Graph learning › Graph self-supervised learning · approach: Brain and biological graph learning · 5 citations (OpenAlex)

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 changedWhereMeasuredEffect
add NRAE-L loss instead of plain AEAE · sin(x) manifoldMSE x100: 1.90 ± 0.23 -> 0.29 ± 0.06Improved
use NRAE-L instead of DAE for denoisingDAE · CIFAR10 noise 0.1PSNR: 20.42 -> 22.21Improved
use NRAE-Q instead of AE on small CIFAR10AE · CIFAR10 Stest MSE: 0.01466 -> 0.00691Improved
use NRAE-L instead of AE on small MNISTAE · MNIST Stest MSE: 0.01002 -> 0.00953Improved
use NRAE-L instead of CAE on small CIFAR100CAE · CIFAR100 Stest MSE: 0.01369 -> 0.00765Improved
combine NRAE-L loss with VAEVAE · MNIST Stest MSE: 0.01091 -> 0.01089No clear effect
combine NRAE-L loss with VAEVAE · CIFAR10 Stest MSE: 0.01620 -> 0.01613No clear effect
combine NRAE-L loss with WAEWAE · CIFAR10 Stest MSE: 0.01431 -> 0.00707Improved
combine NRAE-L loss with CAECAE · CIFAR10 Stest MSE: 0.01208 -> 0.00718Improved
combine NRAE-L loss with SPAESPAE · CIFAR10 Stest MSE: 0.01504 -> 0.00723Improved
combine NRAE-L loss with GRAEGRAE · CIFAR10 Stest MSE: 0.01452 -> 0.00716Improved
use extended NRAE-L (composition) instead of NRAE-L (ablation)NRAE-L · MNIST Stest MSE: 0.00953 -> 0.01000Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy