AI Paper Atlas

Research map / Image generation / GANs

When Do GANs Replicate? On the Choice of Dataset Size

Qianli Feng, Chenqi Guo, Fabian Benitez-Quiroz, Aleix M. Martinez

ICCV 2021 · Image generation › GANs · approach: GAN latent optimization and evaluation · 48 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
use one subset level (one-shot) instead of all levels to fit f1f1 exponential fit · BigGAN FlowerR2: 0.9739 -> 0.8228Hurt
use one subset level (one-shot) instead of all levels to fit f1f1 exponential fit · BigGAN CelebAR2: 0.9388 -> 0.8301Hurt
use one subset level (one-shot) instead of all levels to fit f1f1 exponential fit · StyleGAN2 CelebAR2: 0.9965 -> 0.9958No clear effect
use one subset level (one-shot) instead of all levels to fit f1f1 exponential fit · BigGAN LSUN-bedroomR2: 0.8612 -> 0.8830Improved
add a second subset level (two-shot) to the one-shot fitf1 exponential fit · BigGAN FlowerR2: 0.8228 -> 0.8230No clear effect
add a second subset level (two-shot) to the one-shot fitf1 exponential fit · BigGAN CelebAR2: 0.8301 -> 0.8174Hurt
use one-shot fit instead of full-subset fitf1 exponential fit · BigGAN CelebAMAE_f1 (%): 5.1180 -> 12.0495Hurt
use one-shot fit instead of full-subset fit (median over combinations)f1 exponential fit · All combinationsMAE_f1 (%): 2.6840 -> 2.4215No clear effect
predict replication from dataset size via change of variable instead of from IDf1/f2 fits · All combinations (one-shot)MAE (%): 2.4215 -> 4.5558Hurt
use one-shot fit instead of full fit for predicting dataset sizef2 inverse · BigGAN LSUN-bedroomMAE_f2^-1 (# samples): 2.4715e6 -> 1.4775e7Hurt
increase training dataset sizeBigGAN and StyleGAN2 · CelebA, Flower, LSUN-bedroomreplication percentage: decay factor a in 0.96 to 0.98Improved

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