AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

AVT: Unsupervised Learning of Transformation Equivariant Representations by Autoencoding Variational Transformations

Guo-Jun Qi, Liheng Zhang, Chang Wen Chen, Qi Tian

ICCV 2019 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Structured latent VAEs · 47 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
replace AET transformation prediction with AVT variational mutual-information objectiveNIN unsupervised representation + conv classifier · CIFAR-10error rate (%): 7.82 -> 7.75Improved
use AVT instead of AET-project with non-linear classifierNIN · CIFAR-10error rate (%): 9.41 -> 8.96Improved
use AVT instead of AET-project with 1 FC classifierNIN · CIFAR-10 (1 FC)error rate (%): 16.65 -> 16.19Improved
use AVT instead of AET-project with 3 FC classifierNIN · CIFAR-10 (3 FC)error rate (%): 9.92 -> 9.55Improved
use AVT instead of AET-project for KNN with K=5NIN features · CIFAR-10 KNNerror rate (%): 22.40 -> 21.62Improved
use AVT instead of AET-project for KNN with K=10NIN features · CIFAR-10 KNNerror rate (%): 22.39 -> 23.7Hurt
use AVT instead of AET-project for KNN with K=20NIN features · CIFAR-10 KNNerror rate (%): 23.73 -> 21.51Improved
use AVT unsupervised pretraining instead of fully supervised training with 1000 labels per classNIN conv classifier · CIFAR-10 (limited labels)error rate (%): 16.53 -> 12.27Improved
use AVT unsupervised pretraining instead of AET-project with 20 labels per classNIN conv classifier · CIFAR-10 (20 labels per class)error rate (%): 34.83 -> 35.44Hurt
use the 13-layer network instead of NINAVT + conv · CIFAR-10 (20 labels per class)error rate (%): 35.44 -> 26.2Improved
use AVT instead of AET-project, non-linear classifier on Conv4AlexNet · ImageNettop-1 accuracy (%): 53.2 -> 54.2Improved
use AVT instead of AET-project, non-linear classifier on Conv5AlexNet · ImageNettop-1 accuracy (%): 47.0 -> 48.4Improved

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