Research map / Architectures and efficiency / RNNs and normalization
Infinite Recommendation Networks: A Data-Centric Approach
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 |
|---|---|---|---|
| Use an infinitely-wide autoencoder instead of the strongest listed baseline | ∞-AE · ML-1M | HR@10: 28.54 -> 31.51 | Helped |
| Train ∞-AE on a 500-user Distill-CF summary instead of the full dataset | ∞-AE · Amazon Magazine | HR@10: 27.09 -> 28.27 | Helped |
| Train ∞-AE on a 500-user Distill-CF summary instead of the full dataset | ∞-AE · ML-1M | HR@10: 31.51 -> 31.16 | Hurt |
| Train ∞-AE on a 500-user Distill-CF summary instead of the full dataset | ∞-AE · Douban | HR@10: 23.56 -> 22.98 | Hurt |
| Train ∞-AE on a 500-user Distill-CF summary instead of the full dataset | ∞-AE · Netflix | AUC: 0.9663* -> 0.9728 | Helped |
| Increase ∞-AE depth beyond one or two layers (ablation) | ∞-AE · Recommendation datasets | Performance | Hurt |
| Use Distill-CF summaries rather than heuristic sampling for ∞-AE training | ∞-AE · Recommendation datasets | HR@10: 96-105% of ∞-AE's performance on the full dataset | Helped |
| Train EASE on a Distill-CF summary rather than the full noisy dataset | EASE · ML-1M with 5% noise | HR@10 | Helped |
| Increase noise in the input data before sampling | EASE · ML-1M | HR@10 | Hurt |
| Use Distill-CF summaries rather than generic sampling under injected noise | Recommendation models · ML-1M | HR@10 | Helped |
| Train MVAE on Distill-CF summaries rather than the full dataset | MVAE · ML-1M | Performance | Helped |
| Use Distill-CF summaries instead of ADER or Joint/Individual period data | MVAE continual learning · MovieLens-1M continual learning | Performance | Helped |
Closest papers by meaning
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (ICML 2024 · similarity 0.89)
- KuaiRecV2: Benchmarking Large-Scale Continual Learning for Diversified and Multi-task Recommendation (NeurIPS 2026 · similarity 0.89)
- Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework (AAAI 2024 · similarity 0.88)
- LIME: Link-based User-item Interaction Modeling with Decoupled XOR Attention for Efficient Test Time Scaling (NeurIPS 2026 · similarity 0.88)
- RecFlow: An Industrial Full Flow Recommendation Dataset (ICLR 2025 · similarity 0.87)
- Influential Exemplar Replay for Incremental Learning in Recommender Systems (AAAI 2024 · similarity 0.87)
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.