Research map / Image generation / Editing and style transfer
Sym-Parameterized Dynamic Inference for Mixed-Domain Image Translation
Simyung Chang, SeongUk Park, John Yang, Nojun Kwak
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 |
|---|---|---|---|
| train one sym-parameter network instead of separate hyper-parameter models | 1D toy MLP · 1D toy problem, weights (0.50, 0.50) | weighted loss L: 0.0815 -> 0.0824 | No clear effect |
| train one sym-parameter network instead of separate hyper-parameter models | 1D toy MLP · 1D toy problem, weights (0.00, 1.00) | weighted loss L: 0.0013 -> 0.0062 | Hurt |
| train one sym-parameter network instead of separate hyper-parameter models | SGN Model 1 · Image translation, weights (0.5, 0.5, 0.0) | L_G: 0.407 -> 0.305 | Improved |
| train one sym-parameter network instead of separate hyper-parameter models | SGN Model 1 · Image translation, weights (0.0, 1.0, 0.0) | L_G: 0.308 -> 0.211 | Improved |
| train one sym-parameter network instead of separate hyper-parameter models | SGN Model 1 · Image translation, weights (0.0, 0.5, 0.5) | L_G: 0.205 -> 0.171 | Improved |
| train one sym-parameter network instead of separate hyper-parameter models | SGN Model 1 · Image translation, weights (0.0, 0.0, 1.0) | L_G: 0.015 -> 0.040 | Hurt |
| train one sym-parameter network instead of separate hyper-parameter models | SGN Model 1 · Image translation, weights (1.0, 0.0, 0.0) | L_G: 0.159 -> 0.164 | No clear effect |
| vary loss weights during training (sample sym-parameter) instead of fixing them to (0.5, 0.5) while feeding S as input (ablation) | 1D toy MLP · 1D toy problem | loss | Improved |
| inject sym-parameter with CCAM instead of CONCAT (ablation) | SGN Model 2 · Mixed-domain image translation | qualitative image quality | Improved |
| inject sym-parameter with CCAM instead of CBN (ablation) | SGN Model 2 · Mixed-domain image translation | qualitative image quality | Improved |
| reduce the number of filters in the sym-parameter network (ablation) | 1D toy MLP · 1D toy problem | loss | Hurt |
Closest papers by meaning
- Image-to-image translation for cross-domain disentanglement (NeurIPS 2018 · similarity 0.86)
- Fix the Noise: Disentangling Source Feature for Controllable Domain Translation (CVPR 2023 · similarity 0.86)
- Domain Adaptive Image-to-Image Translation (CVPR 2020 · similarity 0.86)
- Domain-Scalable Unpaired Image Translation via Latent Space Anchoring (TPAMI 2023 · similarity 0.86)
- Unsupervised Image-to-Image Translation Using Domain-Specific Variational Information Bound (NeurIPS 2018 · similarity 0.86)
- Diverse Image-to-Image Translation via Disentangled Representations (ECCV 2018 · similarity 0.86)
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.
