Research map / Optimization / Implicit bias and generalization
Rethinking Neural vs. Matrix-Factorization Collaborative Filtering: the Theoretical Perspectives
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 |
|---|---|---|---|
| take embedding width to infinity | MCF and NCF · Collaborative filtering kernel | limiting NTK | Helped |
| use MCF instead of NCF | Collaborative filtering classifiers · Separable training data | implicit max-margin solution | Mixed |
| use log or exponential loss with homogeneous predictors on separable data | MCF and NCF · Training data | normalized margin solution | Helped |
| use a wider network and its linearized predictor | MCF and NCF · Finite-time gradient flow | decision-boundary discrepancy | Helped |
| use learning rates from the tested range | MCF and NCF · Inductive CF experiments | converged performance | No clear effect |
| use NCF with addition instead of concatenation | NCF · Inductive CF experiments | performance pattern | No clear effect |
Closest papers by meaning
- Towards Open-World Recommendation: An Inductive Model-based Collaborative Filtering Approach (ICML 2021 · similarity 0.89)
- Bayesian Deep Collaborative Matrix Factorization (AAAI 2019 · similarity 0.89)
- Recommendations with Sparse Comparison Data: Provably Fast Convergence for Nonconvex Matrix Factorization (ICML 2025 · similarity 0.88)
- Learning-Efficient Yet Generalizable Collaborative Filtering for Item Recommendation (ICML 2024 · similarity 0.87)
- How Does Message Passing Improve Collaborative Filtering? (NeurIPS 2024 · similarity 0.87)
- DR-VAE: Debiased and Representation-enhanced Variational Autoencoder for Collaborative Recommendation (AAAI 2025 · 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.