Reproduce “Hypercomplex Prompt-aware Multimodal Recommendation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zheyu Chen, Jinfeng Xu, Hewei Wang, Shuo Yang, Zitong Wan, Haibo Hu
A framework called HPMRec, developed by researchers from several universities including The Hong Kong Polytechnic University and Carnegie Mellon University, enhances multimodal recommender systems by combining hypercomplex embeddings, prompt-aware compensation, and nonlinear fusion strategies. This approach achieves state-of-the-art recommendation performance on four Amazon datasets, consistently outperforming existing baselines in Recall@K and NDCG@K metrics.
Key insights
- Multi-component hypercomplex embeddings provide significantly richer and more diverse representations for multimodal features compared to single-vector embeddings.
- Nonlinear hypercomplex multiplication is an effective strategy for capturing deep, intricate semantic correlations between different modalities, improving fusion.
- Learnable prompts can dynamically compensate for feature misalignment in multi-component embeddings and concurrently alleviate GCN over-smoothing.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.