Research map / Image generation / GANs
Feedback Adversarial Learning: Spatial Feedback for Improving Generative Adversarial Networks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add discriminator spatial feedback and previous-generation conditioning | GAN · CelebA | FID: 22.56 -> 18.52 | Helped |
| Use feedback at generation step 1 | GAN · CelebA | FID: 22.56 -> 26.49 | Hurt |
| Continue feedback refinement from step 1 to step 3 | GAN · CelebA | FID: 26.49 -> 18.52 | Helped |
| Add feedback and refine to step 3 | Pix2Pix · Cityscapes | Cat IOU: 0.380 -> 0.418 | Helped |
| Add feedback and refine to step 3 | Pix2Pix · Cityscapes | Cls IOU: 0.655 -> 0.692 | Helped |
| Add feedback and refine to step 3 | Pix2Pix · Cityscapes | LPIPS (Val): 0.428 -> 0.429 | Hurt |
| Add feedback and refine to step 3 | Pix2Pix · Cityscapes | LPIPS (Train): 0.320 -> 0.254 | Helped |
| Add feedback and refine to step 3 | Pix2Pix · NYU-Depth-V2 | REL: 0.191 -> 0.181 | Helped |
| Add feedback and refine to step 3 | Pix2Pix · NYU-Depth-V2 | δ1: 0.892 -> 0.701 | Hurt |
| Add feedback and refine to step 3 | Pix2Pix · NYU-Depth-V2 | δ2: 0.961 -> 0.908 | Hurt |
| Add discriminator spatial feedback and recurrent refinement | VoxelGAN · ShapeNet airplane | Classification accuracy: 93.0% -> 95.6% | Helped |
| Add discriminator spatial feedback and recurrent refinement | VoxelGAN · ShapeNet car | Classification accuracy: 98.1% -> 99.1% | Helped |
Closest papers by meaning
- Neural Networks with Recurrent Generative Feedback (NeurIPS 2020 · similarity 0.88)
- Generalized Adversarially Learned Inference (AAAI 2021 · similarity 0.86)
- Progressive Growing of GANs for Improved Quality, Stability, and Variation (ICLR 2018 · similarity 0.86)
- Towards Principled Methods for Training Generative Adversarial Networks (ICLR 2017 · similarity 0.86)
- Improving Generative Adversarial Networks with Denoising Feature Matching (ICLR 2017 · similarity 0.86)
- DO-GAN: A Double Oracle Framework for Generative Adversarial Networks (CVPR 2022 · similarity 0.85)
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.