AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

On Positive-Unlabeled Classification in GAN

Tianyu Guo, Chang Xu, Jiajun Huang, Yunhe Wang, Boxin Shi, Chao Xu, Dacheng Tao

CVPR 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 54 citations (OpenAlex)

Check 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
treat generated samples as unlabeled and train the discriminator to distinguish high- from low-quality generated samplesGANs on MNIST, Fashion, CIFAR-10, CelebA-64, and CAT-64 · MNIST, Fashion, CIFAR-10, CelebA-64, CAT-64FID: 27.64 -> 22.32Improved
use PU learning rather than ordinary real-versus-generated classificationGANs · MNISTFID: 13.47 -> 13.96Hurt
use PU learning rather than ordinary real-versus-generated classificationGANs · CAT-64 stabilityFID mean: 48.36 -> 16.38Improved
increase the learning rate to .001 (ablation)PUSGAN · CIFAR-10FID: 31.02 -> 40.93Hurt
remove batch normalization (ablation)PUSGAN · CIFAR-10FID: 31.02 -> 37.85Hurt
replace activation functions with Tanh (ablation)PUSGAN · CIFAR-10FID: 31.02 -> 54.70Hurt
increase the learning rate to .001 (ablation)PULSGAN · CIFAR-10FID: 22.32 -> 37.65Hurt
remove batch normalization (ablation)PULSGAN · CIFAR-10FID: 22.32 -> 37.34Hurt
replace activation functions with Tanh (ablation)PULSGAN · CIFAR-10FID: 22.32 -> 51.45Hurt
start the class prior at 0.3 and increase it by 0.1 every 10k iterations (ablation)PUSGAN · CAT-64FIDHurt
fix the class prior at 0.3 (ablation)PUSGAN · CAT-64FIDMixed
fix the class prior at 0.5 (ablation)PUSGAN · CAT-64FIDMixed

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