Research map / Image generation / GANs
Sampling-Decomposable Generative Adversarial Recommender
Binbin Jin, Defu Lian, Zheng Liu, Qi Liu, Jianhui Ma, Xing Xie, Enhong Chen
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 |
|---|---|---|---|
| Use self-normalized importance sampling to compensate for the limited-capacity generator | SD-GAR · CiteULike | NDCG@50: 0.1269 -> 0.1365 | Improved |
| Use self-normalized importance sampling to compensate for the limited-capacity generator | SD-GAR · Gowalla | NDCG@50: 0.1490 -> 0.1620 | Improved |
| Use self-normalized importance sampling to compensate for the limited-capacity generator | SD-GAR · MovieLens | NDCG@50: 0.2858 -> 0.3134 | Improved |
| Use self-normalized importance sampling to compensate for the limited-capacity generator | SD-GAR · Amazon | NDCG@50: 0.0580 -> 0.0691 | Improved |
| Use self-normalized importance sampling to compensate for the limited-capacity generator | SD-GAR · Echonest | NDCG@50: 0.1138 -> 0.1267 | Improved |
| Replace SD-GAR's learned generator with a uniform distribution and disable gradients through sample importance (ablation) | SD-GAR · CiteULike, Gowalla, Echonest | NDCG@50: 0.1365 -> 0.1269; 0.1620 -> 0.1490; 0.1267 -> 0.1138 | Hurt |
| Use a sampling-decomposable generator with Vose-Alias sampling | SD-GAR vs IRGAN · Amazon | discriminator training speedup: 20x | Improved |
| Use a sampling-decomposable generator with Vose-Alias sampling | SD-GAR vs IRGAN · Echonest | discriminator training speedup: 10x | Improved |
| Use a sampling-decomposable generator | SD-GAR vs IRGAN · Amazon and Echonest | generator training speedup: 5x | Improved |
| Increase the number of discriminator negative samples beyond 5 (ablation) | SD-GAR · CiteULike and Gowalla | NDCG@50 | Mixed |
| Vary the number of item and context samples used to learn the generator (ablation) | SD-GAR · CiteULike and Gowalla | NDCG@50 | No clear effect |
| Increase the embedding size (ablation) | SD-GAR · CiteULike and Gowalla | NDCG@50 | Improved |
Closest papers by meaning
- Adversarial Binary Collaborative Filtering for Implicit Feedback (AAAI 2019 · similarity 0.88)
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (ICML 2024 · similarity 0.86)
- Collaborative Sampling in Generative Adversarial Networks (AAAI 2020 · similarity 0.85)
- Adversarial Counterfactual Learning and Evaluation for Recommender System (NeurIPS 2020 · similarity 0.85)
- Sparse Enhanced Network: An Adversarial Generation Method for Robust Augmentation in Sequential Recommendation (AAAI 2024 · similarity 0.83)
- Echoes in Filter Bubble: Diagnosing and Curing Popularity Bias in Generative Recommender Systems (NeurIPS 2026 · similarity 0.83)
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.
