Research map / Probabilistic and scientific modeling / Molecules and proteins
Learning to Recommend from Sparse Data via Generative User Feedback
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the CF-SFL synthetic feedback loop to VAE (ablation) | VAE recommender · ML-20M | NDCG@100: 0.42546 -> 0.43472; +9.26 | Helped |
| Add the CF-SFL synthetic feedback loop to WARP (ablation) | WARP recommender · ML-20M | NDCG@100: 0.31228 -> 0.33987; +27.59 | Helped |
| Add the CF-SFL synthetic feedback loop to MF (ablation) | MF recommender · ML-20M | NDCG@100: 0.41587 -> 0.41902; +3.15 | Helped |
| Add the CF-SFL synthetic feedback loop to DAE (ablation) | DAE recommender · ML-20M | NDCG@100: 0.42056 -> 0.42307; +2.51 | Helped |
| Add the CF-SFL synthetic feedback loop to VAE-(Gaussian) (ablation) | VAE-(Gaussian) recommender · ML-20M | NDCG@100: 0.42019 -> 0.42751; +7.32 | Helped |
| Add the CF-SFL synthetic feedback loop to VAE-(β=0) (ablation) | VAE-(β=0) recommender · ML-20M | NDCG@100: 0.42027 -> 0.42539; +5.02 | Helped |
| Add the CF-SFL synthetic feedback loop to VAE-Linear (ablation) | VAE-Linear recommender · ML-20M | NDCG@100: 0.41563 -> 0.41597; +0.34 | Helped |
| Add a reward estimator without the feedback generator and recurrent feedback loop (ablation) | VAE · ML-20M, Netflix, and MSD | Recommendation metrics | No clear effect |
| Increase unrolling steps within the tested range (ablation) | CF-SFL · ML-20M, Netflix, and MSD | NDCG@100 | Helped |
| Use CF-SFL rather than the VAE* baseline | VAE recommender · ML-20M | Recall@20: 0.395 -> 0.404 | Helped |
| Use CF-SFL rather than the VAE* baseline | VAE recommender · Netflix | NDCG@100: 0.386 -> 0.394 | Helped |
| Use CF-SFL rather than the VAE* baseline | VAE recommender · MSD | Recall@50: 0.356 -> 0.369 | Helped |
Closest papers by meaning
- LLM-Powered User Simulator for Recommender System (AAAI 2025 · similarity 0.88)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.87)
- Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework (AAAI 2024 · similarity 0.87)
- Recommendations with Sparse Comparison Data: Provably Fast Convergence for Nonconvex Matrix Factorization (ICML 2025 · similarity 0.86)
- MTRec: Learning to Align with User Preferences via Mental Reward Models (NeurIPS 2025 · similarity 0.86)
- Influential Exemplar Replay for Incremental Learning in Recommender Systems (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.