Research map / Vision-language and multimodal / Human and embodied modeling
Learning Long- and Short-Term User Literal-Preference with Multimodal Hierarchical Transformer Network for Personalized Image Caption
Wei Zhang, Yue Ying, Pan Lu, Hongyuan Zha
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 |
|---|---|---|---|
| hierarchical transformer with short-term user encoder over recent captions plus long-term user ID | EICP · Instagram | CIDEr: 0.094 -> 0.125 | Improved |
| MHTN vs EICP | EICP · YFCC100M | CIDEr: 0.263 -> 0.408 | Improved |
| add temporal embedding (ablation) | MHTN · Instagram | CIDEr: 0.119 -> 0.125 | Improved |
| add target image input (ablation) | MHTN · Instagram | CIDEr: 0.087 -> 0.125 | Improved |
| add long-term user representation (ablation) | MHTN · YFCC100M | CIDEr: 0.297 -> 0.408 | Improved |
| add short-term user encoder (ablation) | MHTN · YFCC100M | CIDEr: 0.332 -> 0.408 | Improved |
| use transformer instead of time-aware RNN as short-term encoder (ablation) | MHTN · YFCC100M | CIDEr: 0.364 -> 0.408 | Improved |
| increase number of recent posts from 5 to 10 (ablation) | MHTN · Instagram | CIDEr: 0.125 -> 0.126 | No clear effect |
| increase number of recent posts without temporal embedding (ablation) | MHTN w/o Temporal Emb. · Instagram | CIDEr: 0.119 -> 0.110 (2 to 10 posts) | Hurt |
Closest papers by meaning
- Yo'Chameleon: Personalized Vision and Language Generation (CVPR 2025 · similarity 0.86)
- Tailored Visions: Enhancing Text-to-Image Generation with Personalized Prompt Rewriting (CVPR 2024 · similarity 0.86)
- MM-R1: Unleashing the Power of Unified Multimodal Large Language Models for Personalized Image Generation (AAAI 2026 · similarity 0.86)
- MM-SeR: Multimodal Self-Refinement for Lightweight Image Captioning (CVPR 2026 · similarity 0.85)
- Revisit Large-Scale Image-Caption Data in Pre-training Multimodal Foundation Models (ICLR 2025 · similarity 0.85)
- RePIC: Reinforced Post-Training for Personalizing Multi-Modal Language Models (NeurIPS 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.
