Research map / Label-efficient and robust learning / Domain adaptation and generalization
Learning to Transfer: Unsupervised Domain Translation via Meta-Learning
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 |
|---|---|---|---|
| meta-learned initialization (MT-GAN) instead of training CycleGAN from scratch | CycleGAN · Face identity translation 5-shot | Top-1 accuracy: 11.21 -> 13.05 | Helped |
| meta-learned initialization (MT-GAN) instead of training CycleGAN from scratch | CycleGAN · Face identity translation 10-shot | Top-1 accuracy: 19.04 -> 21.12 | Helped |
| meta-learned initialization instead of CycleGAN training | CycleGAN · labels->photos 10-shot | FID (x10): 28.69 -> 26.08 | Helped |
| meta-learned initialization instead of CycleGAN training | CycleGAN · photos<-maps 10-shot | FID (x10): 25.57 -> 19.64 | Helped |
| meta-learned initialization instead of CycleGAN training | CycleGAN · painting<->photo tasks 10-shot | FID (x10): 19.26 -> 18.42 | Helped |
| increase shots from 10 to 25 (gap between MT-GAN and CycleGAN) | MT-GAN · Face identity translation | Top-1 accuracy improvement over CycleGAN: 2.08% -> 0.41% | Hurt |
| meta-learned initialization vs StarGAN | StarGAN · Face identity translation 5-shot | Top-1 accuracy: 4.36 -> 13.05 | Helped |
Closest papers by meaning
- DINO: A Conditional Energy-Based GAN for Domain Translation (ICLR 2021 · similarity 0.90)
- Unsupervised Image-to-Image Translation Networks (NeurIPS 2017 · similarity 0.89)
- Unsupervised Image-to-Image Translation With Generative Prior (CVPR 2022 · similarity 0.88)
- Go From the General to the Particular: Multi-Domain Translation with Domain Transformation Networks (AAAI 2020 · similarity 0.88)
- Conditional Image-to-Image Translation (CVPR 2018 · similarity 0.88)
- Towards Identifiable Unsupervised Domain Translation: A Diversified Distribution Matching Approach (ICLR 2024 · 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.