Research map / Image generation / Autoregressive and latent generation
FIANCEE: Faster Inference of Adversarial Networks via Conditional Early Exits
Read the paperCheck 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 |
|---|---|---|---|
| Route easier inputs through early exits rather than always using the full backbone | FIANCEE on OASIS · Cityscapes | GFLOPs: 120 -> 319 | Helped |
| Route easier inputs through early exits rather than always using the full backbone | FIANCEE on MegaPortraits · VoxCeleb2 | GFLOPs: 47 -> 213 | Helped |
| Add database guidance to early exits (ablation) | OASIS Branch 1 (SF=1/4) · Cityscapes | FID: 69.6 -> 54.9 | Helped |
| Add database guidance to early exits (ablation) | OASIS Branch 1 (SF=1/4) · Cityscapes | mIOU: 57.5 -> 65.5 | Helped |
| Add database guidance to the earliest exit (ablation) | MegaPortraits Branch 1 (SF=1/15) · VoxCeleb2 cross-reenactment | FID: 66.87 -> 57.25 | Helped |
| Use database guidance on later MegaPortraits exits (SF=1/3) (ablation) | MegaPortraits · VoxCeleb2 cross-reenactment | FID: 49.08 -> 50.44 | Mixed |
| Vary branch width using smaller scale factors (ablation) | FIANCEE branches · Cityscapes and VoxCeleb2 | GFLOPs | Helped |
| Use a predictor to select an exit per input instead of relying on one fixed exit (ablation) | FIANCEE · Cityscapes and VoxCeleb2 | LPIPS threshold consistency | Helped |
Closest papers by meaning
- Large Scale GAN Training for High Fidelity Natural Image Synthesis (ICLR 2019 · similarity 0.89)
- Learning Semantic-aware Normalization for Generative Adversarial Networks (NeurIPS 2020 · similarity 0.88)
- Improving Generative Adversarial Networks with Denoising Feature Matching (ICLR 2017 · similarity 0.88)
- TR0N: Translator Networks for 0-Shot Plug-and-Play Conditional Generation (ICML 2023 · similarity 0.88)
- Cluster-Guided Image Synthesis With Unconditional Models (CVPR 2022 · similarity 0.88)
- Unlocking Pre-trained Image Backbones for Semantic Image Synthesis (CVPR 2024 · similarity 0.88)
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.