Research map / Label-efficient and robust learning / Domain adaptation and generalization
Leveraging Distribution Alignment via Stein Path for Cross-Domain Cold-Start Recommendation
Weiming Liu, Jiajie Su, Chaochao Chen, Xiaolin Zheng
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 Stein path distribution alignment between warm and cold auxiliary embeddings (SP(I)) to base (ablation) | DisAlign-Base · Douban Movie->Book | HR@20: .2991 -> .3375 | Improved |
| add Stein path distribution alignment between warm and cold auxiliary embeddings (SP(I)) to base (ablation) | DisAlign-Base · Douban Movie->Music | HR@20: .3258 -> .3650 | Improved |
| additionally align cold embeddings with warm collaborative embeddings V (SP vs SP(I)) (ablation) | DisAlign-SP(I) · Douban Movie->Book | HR@20: .3375 -> .3428 | Improved |
| additionally align cold embeddings with V (SP vs SP(I)) (ablation) | DisAlign-SP(I) · Douban Book->Movie | HR@20: .3023 -> .3028 | No clear effect |
| use proxy Stein path instead of full Stein path (ablation) | DisAlign-SP · Douban Movie->Music | HR@20: .3734 -> .3795 | Improved |
| use proxy Stein path instead of full Stein path (ablation) | DisAlign-SP · Douban Movie->Book | HR@20: .3428 -> .3401 | Hurt |
| use proxy Stein path instead of full Stein path (ablation) | DisAlign-SP · per-epoch training time | time per epoch | Improved |
| use Stein path alignment instead of ESAM correlation alignment | ESAM · Douban Movie->Book | HR@20: .3146 -> .3428 | Improved |
| use Stein path alignment instead of DARec adversarial alignment | DARec · Amazon Movie->Music | HR@20: .2981 -> .3155 | Improved |
| increase alignment loss weight lambda too high (ablation) | DisAlign · hyperparameter sensitivity | recommendation accuracy | Hurt |
| increase embedding dimension D (ablation) | DisAlign-SP/PSP · hyperparameter sensitivity | recommendation accuracy | Improved |
Closest papers by meaning
- Domain-Level Disentanglement Framework Based on Information Enhancement for Cross-Domain Cold-Start Recommendation (AAAI 2025 · similarity 0.90)
- Learning Accurate and Bidirectional Transformation via Dynamic Embedding Transportation for Cross-Domain Recommendation (AAAI 2024 · similarity 0.88)
- DisCo: Graph-Based Disentangled Contrastive Learning for Cold-Start Cross-Domain Recommendation (AAAI 2025 · similarity 0.87)
- Towards Unbiased Information Extraction and Adaptation in Cross-Domain Recommendation (AAAI 2025 · similarity 0.87)
- Reducing Item Discrepancy via Differentially Private Robust Embedding Alignment for Privacy-Preserving Cross Domain Recommendation (ICML 2024 · similarity 0.87)
- Graph Disentangled Contrastive Learning with Personalized Transfer for Cross-Domain Recommendation (AAAI 2024 · similarity 0.86)
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.
