Research map / Vision-language and multimodal / Cross-modal learning
Multi-Aspect Cross-modal Quantization for Generative Recommendation
Fuwei Zhang, Xiaoyu Liu, Dongbo Xi, Jishen Yin, Huan Chen, Peng Yan, Fuzhen Zhuang, Zhao Zhang
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 |
|---|---|---|---|
| Add cross-modal contrastive loss in RQ layers (ablation) | MACRec · Instruments | HR@10: 0.1289 -> 0.1363 | Improved |
| Add cross-modal contrastive loss in RQ layers (ablation) | MACRec · Games | HR@10: 0.1018 -> 0.1078 | Improved |
| Add reconstruction alignment loss (ablation) | MACRec · Games | HR@10: 0.1026 -> 0.1078 | Improved |
| Add reconstruction alignment loss (ablation) | MACRec · Arts | HR@10: 0.1301 -> 0.1329 | Improved |
| Add implicit alignment (ablation) | MACRec · Instruments | HR@10: 0.1312 -> 0.1363 | Improved |
| Add explicit alignment tasks (ablation) | MACRec · Arts | HR@10: 0.1299 -> 0.1329 | Improved |
| Use MACRec instead of MQL4GRec | MACRec · Instruments | HR@10: 0.1291 -> 0.1363 | Improved |
| Use MACRec instead of MISSRec on Games HR@5 | MACRec · Games | HR@5: 0.0674 -> 0.0671 | No clear effect |
| Use MACRec instead of MISSRec on Instruments HR@10 | MACRec · Instruments | HR@10: 0.1361 -> 0.1363 | No clear effect |
| Use cross-modal quantization to reduce item collisions | MACRec · Arts text IDs | collision rate (%): 5.15 -> 4.24 | Improved |
| Apply cross-modal contrastive loss from the first RQ layer instead of the third (ablation) | MACRec · Instruments | HR@10 | Hurt |
Closest papers by meaning
- Multimodal Quantitative Language for Generative Recommendation (ICLR 2025 · similarity 0.91)
- Towards Unified Multi-Modal Personalization: Large Vision-Language Models for Generative Recommendation and Beyond (ICLR 2024 · similarity 0.89)
- Rethinking Semantic ID Construction for Generative Recommendation: SimHash with Parallel Decoding and Semantic Alignment (NeurIPS 2026 · similarity 0.89)
- MENTOR: Multi-level Self-supervised Learning for Multimodal Recommendation (AAAI 2025 · similarity 0.88)
- Curriculum Conditioned Diffusion for Multimodal Recommendation (AAAI 2025 · similarity 0.88)
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (ICML 2024 · similarity 0.88)
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.
