Research map / Image generation / Text-to-image diffusion
Learning Personalized Photographic Style from Pairwise User Preferences
Jinwoo Kim, Jihye Yoo, Seon Joo Kim
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 a shared decoder conditioned on user preference embeddings instead of per-user decoders | PPS baselines · PPSD validation | CQS (Delta E00): 0.154 -> 0.204 | Improved |
| use UPE model instead of exemplar-based inference | PPS baselines · PPSD validation | PSNR vs preferred: 25.52 -> 36.67 | Improved |
| increase number of training users from 50 to 200 to 471 (ablation) | User Preference Embedding · PPSD validation | CQS (Delta E00): 0.185 -> 0.192 -> 0.204 | Improved |
| increase number of training users from 50 to 471 (ablation) | User Preference Embedding · PPSD validation | PSNR vs preferred: 30.4 -> 36.7 | Improved |
| increase reference samples from 4 to 16 (ablation) | User Preference Embedding · PPSD validation | CQS (Delta E00): 0.195 -> 0.204 | Improved |
| increase reference samples from 16 to 32 (ablation) | User Preference Embedding · PPSD validation | CQS (Delta E00): 0.204 -> 0.195 | Hurt |
| change reference selection strategy at N=16 (ablation) | User Preference Embedding · PPSD validation | CQS (Delta E00): 0.204 / 0.202 / 0.203 | No clear effect |
| train on both preferred and non-preferred images with curriculum on weight w | all three PPS baselines · PPSD validation | training objective | No clear effect |
Closest papers by meaning
- Personalized Image Aesthetics (ICCV 2017 · similarity 0.91)
- Correct for Whom? Subjectivity and the Evaluation of Personalized Image Aesthetics Assessment Models (AAAI 2023 · similarity 0.88)
- PreferThinker: Reasoning-based Personalized Image Preference Assessment (ICLR 2026 · similarity 0.88)
- Personalized Image Editing in Text-to-Image Diffusion Models via Collaborative Direct Preference Optimization (NeurIPS 2025 · similarity 0.86)
- Learning Multi-Dimensional Human Preference for Text-to-Image Generation (CVPR 2024 · similarity 0.85)
- Conditional Balance: Improving Multi-Conditioning Trade-Offs in Image Generation (CVPR 2025 · 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.
