Research map / Label-efficient and robust learning / Recognition and long-tail
Symmetric Metric Learning with Adaptive Margin for Recommendation
Mingming Li, Shuai Zhang, Fuqing Zhu, Wanhui Qian, Liangjun Zang, Jizhong Han, Songlin Hu
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 user-specific adaptive margin to base (ablation) | SML (CML-like base) · Instant Video | HR@5: 0.6911 -> 0.7256 | Improved |
| add user-specific adaptive margin to base (ablation) | SML (CML-like base) · IMDB | HR@5: 0.4593 -> 0.4617 | Improved |
| add user-specific adaptive margin to base (ablation) | SML (CML-like base) · Yelp | HR@5: 0.5667 -> 0.5709 | Improved |
| add item-centric symmetric loss to base (ablation) | SML (CML-like base) · Instant Video | HR@5: 0.6911 -> 0.7326 | Improved |
| add item-centric symmetric loss to base (ablation) | SML (CML-like base) · IMDB | HR@5: 0.4593 -> 0.4500 | Hurt |
| add item-specific margin on top of s+u (ablation) | SML (CML-like base) · Instant Video | HR@5: 0.7347 -> 0.7309 | No clear effect |
| add item-specific margin on top of s+u (ablation) | SML (CML-like base) · IMDB | HR@5: 0.4608 -> 0.4827 | Improved |
| add item-specific margin on top of s+u (ablation) | SML (CML-like base) · Yelp | HR@5: 0.5749 -> 0.5900 | Improved |
| increase lambda (item-centric loss weight) above 0.01 | SML · Instant Video, IMDB | HR@K | Hurt |
| set adaptive margin strength gamma too large (>100) | SML · Instant Video | HR@K | Hurt |
| use full SML instead of CML | SML · IMDB | HR@5: 0.4593 -> 0.4827 | Improved |
Closest papers by meaning
- PSL: Rethinking and Improving Softmax Loss from Pairwise Perspective for Recommendation (NeurIPS 2024 · similarity 0.82)
- Learning-Efficient Yet Generalizable Collaborative Filtering for Item Recommendation (ICML 2024 · similarity 0.82)
- From Pairwise to Ranking: Climbing the Ladder to Ideal Collaborative Filtering with Pseudo-Ranking (AAAI 2025 · similarity 0.81)
- Adaptive Hardness Negative Sampling for Collaborative Filtering (AAAI 2024 · similarity 0.81)
- ENSUR: Equitable and Statistically Unbiased Recommendation (ICML 2025 · similarity 0.81)
- ImplicitSLIM and How it Improves Embedding-based Collaborative Filtering (ICLR 2024 · 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.
