Research map / Vision-language and multimodal / Human and embodied modeling
Learning User Representations for Open Vocabulary Image Hashtag Prediction
Check 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 |
|---|---|---|---|
| Add visual information to the user's hashtag history representation (ablation) | User-conditional hashtag prediction · YFCC100M open vocabulary, seen hashtags | A@1: 65.06 -> 67.46 | Helped |
| Add visual information to the user's hashtag history representation (ablation) | User-conditional hashtag prediction · YFCC100M open vocabulary, unseen hashtags | A@1: 44.07 -> 45.98 | Helped |
| Add pretrained GloVe embeddings (ablation) | User-conditional hashtag prediction · YFCC100M open vocabulary, seen hashtags | A@1: 46.24 -> 67.46 | Helped |
| Use a temporal GRU over the user's image-hashtag history rather than pooled hashtag history (ablation) | User representation for fixed-vocabulary hashtag prediction · YFCC100M fixed vocabulary | A@1: 62.09 -> 71.90 | Helped |
| Use a dynamic history instead of a fixed training-image user history (ablation) | User representation in transductive hashtag prediction · YFCC100M fixed vocabulary, transductive users | A@1: 48.20 -> 73.19 | Helped |
| Represent user history with element-wise image-hashtag fusion (ablation) | User representation · YFCC100M fixed vocabulary, fusion analysis | A@1: 65.16 -> 67.28 | Helped |
| Replace element-wise product fusion with TIRG fusion (ablation) | User representation · YFCC100M fixed vocabulary, fusion analysis | A@1: 67.28 -> 63.97 | Hurt |
| Increase user representation dimension (ablation) | User representation model · YFCC100M hashtag prediction | R@10 | Helped |
Closest papers by meaning
- Separating Self-Expression and Visual Content in Hashtag Supervision (CVPR 2018 · similarity 0.90)
- Weakly Supervised Deep Image Hashing Through Tag Embeddings (CVPR 2019 · similarity 0.86)
- An End-To-End Graph Attention Network Hashing for Cross-Modal Retrieval (NeurIPS 2024 · similarity 0.82)
- Vision-guided Text Mining for Unsupervised Cross-modal Hashing with Community Similarity Quantization (AAAI 2025 · similarity 0.82)
- Robust Self-Paced Hashing for Cross-Modal Retrieval with Noisy Labels (AAAI 2025 · similarity 0.82)
- Improving Multimodal Social Media Popularity Prediction via Selective Retrieval Knowledge Augmentation (AAAI 2025 · similarity 0.81)
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.