AI Paper Atlas

Research map / Image generation / GANs

ExGAN: Adversarial Generation of Extreme Samples

Siddharth Bhatia, Arjit Jain, Bryan Hooi

AAAI 2021 · Image generation › GANs · approach: GAN latent optimization and evaluation · 40 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 distribution shifting to focus training on increasingly extreme rainfall patterns (ablation)ExGAN · US precipitation test extremesFID: 0.0493 ± 0.0097 -> 0.0236 ± 0.0037Improved
Use distribution shifting to focus training on increasingly extreme rainfall patterns (ablation)ExGAN · US precipitation test extremesReconstruction Loss: 0.0166 compared to 0.0172No clear effect
Replace DCGAN rejection sampling with EVT-based conditional generation and distribution shiftingUS precipitation sampling at τ=0.01 · US precipitation, τ=0.01; 100 samplessampling time: 7.564s -> 0.002sImproved
Use ExGAN instead of DCGAN rejection samplingUS precipitation sampling at τ=0.05 · US precipitation, τ=0.05; 100 samplessampling time: 1.230s -> 0.002sImproved
Use ExGAN instead of DCGANUS precipitation test extremes · US precipitation test extremesFID: 0.0406 ± 0.0063 -> 0.0236 ± 0.0037Improved
Use ExGAN instead of DCGANUS precipitation test extremes · US precipitation test extremesReconstruction Loss: 0.0292 -> 0.0172Improved
Use a more gradual distribution shift with c=0.90, k=27 instead of c=0.75, k=10 (ablation)ExGAN · US precipitation test extremesFID: 0.0236 ± 0.0037 -> 0.0223 ± 0.0121Improved
Use a more gradual distribution shift with c=0.90, k=27 instead of c=0.75, k=10 (ablation)ExGAN · US precipitation test extremesReconstruction Loss: 0.0172 -> 0.0169Improved
Use c=0.90, k=27 instead of c=0.75, k=10 for distribution shifting (ablation)ExGAN · US precipitation test extremesMAPE: 3.14 ± 3.08 -> 3.05 ± 3.14Improved
Use c=0.24, k=2 instead of c=0.75, k=10 for distribution shifting (ablation)ExGAN · US precipitation test extremesFID: 0.0236 ± 0.0037 -> 0.0367 ± 0.0096Hurt
Use c=0.49, k=4 instead of c=0.75, k=10 for distribution shifting (ablation)ExGAN · US precipitation test extremesFID: 0.0236 ± 0.0037 -> 0.0304 ± 0.0109Hurt

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