Research map / Image generation / GANs
On the Effectiveness of Least Squares Generative Adversarial Networks
Xudong Mao, Qing Li, Haoran Xie, Raymond Y.K. Lau, Zhen Wang, Stephen Paul Smolley
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use LSGANs (-1,1,0) instead of NS-GANs | LSGANs · LSUN-bedroom | FID: 28.04 -> 21.55 | Improved |
| use LSGANs (-1,1,0) instead of NS-GANs | LSGANs · Cat | FID: 15.81 -> 14.28 | Improved |
| use LSGANs (-1,1,0) instead of NS-GANs | LSGANs · CIFAR-10 | FID: 35.25 -> 35.19 | No clear effect |
| use LSGANs (-1,1,0) instead of WGANs-GP | LSGANs · ImageNet | FID: 62.05 -> 68.95 | Hurt |
| use LSGANs (-1,1,0) instead of WGANs-GP | LSGANs · Cat | FID: 29.03 -> 14.28 | Improved |
| use parameters (-1,1,0) instead of (0,1,1) in LSGANs (ablation) | LSGANs · LSUN-bedroom | FID: 27.21 -> 21.55 | Improved |
| use parameters (-1,1,0) instead of (0,1,1) in LSGANs (ablation) | LSGANs · ImageNet | FID: 72.54 -> 68.95 | Improved |
| use parameters (-1,1,0) instead of (0,1,1) in LSGANs (ablation) | LSGANs · SVHN | convergence speed | Improved |
| use LSGANs instead of NS-GANs on Gaussian mixture | LSGANs · 8-Gaussian mixture | runs with mode collapse: 99 -> 5 | Improved |
| use LSGANs instead of NS-GANs | LSGANs · LSUN-bedroom human study | votes: 43.6% -> 56.4% | Improved |
| use LSGANs instead of WGANs-GP for training time | LSGANs · LSUN-bedroom | wall-clock time to FID about 22: 4,600 minutes -> 1,100 minutes | Improved |
| use LSGANs instead of NS-GANs without BN in generator, Adam | LSGANs · LSUN-bedroom | successful runs: 0 of 10 -> 5 of 10 | Improved |
Closest papers by meaning
- Least Squares Generative Adversarial Networks (ICCV 2017 · similarity 0.98)
- TRAINING GENERATIVE ADVERSARIAL NETWORKS VIA PRIMAL-DUAL SUBGRADIENT METHODS: A LAGRANGIAN PERSPECTIVE ON GAN (ICLR 2018 · similarity 0.91)
- SGAN: An Alternative Training of Generative Adversarial Networks (CVPR 2018 · similarity 0.90)
- Training Generative Adversarial Networks with Limited Data (NeurIPS 2020 · similarity 0.90)
- Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities (IJCV 2019 · similarity 0.90)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.89)
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.
