Research map / Learning theory and algorithms / Kernels and regression theory
Information-Theoretic Understanding of Population Risk Improvement with Model Compression
Yuheng Bu, Weihao Gao, Shaofeng Zou, Venugopal Veeravalli
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 |
|---|---|---|---|
| Compress the ERM solution using K-means with fewer clusters | Linear regression ERM solution · Linear regression | population risk | Improved |
| Add diameter regularization to Hessian-weighted K-means | Hessian-weighted K-means · MNIST | population risk | Improved |
| Add diameter regularization to Hessian-weighted K-means | Hessian-weighted K-means · CIFAR10 | population risk | Improved |
| Increase the diameter-regularization strength β (ablation) | Diameter-regularized Hessian-weighted K-means · MNIST, K=7 | generalization error | Improved |
| Increase the diameter-regularization strength β (ablation) | Diameter-regularized Hessian-weighted K-means · MNIST, K=7 | empirical risk distortion | Hurt |
| Increase the diameter-regularization strength β (ablation) | Diameter-regularized Hessian-weighted K-means · MNIST, K=7 | population risk | Mixed |
Closest papers by meaning
- Compression, Generalization and Learning (JMLR 2023 · similarity 0.88)
- Does Compression Imply Generalization? A Minimum Description Length Perspective (NeurIPS 2026 · similarity 0.83)
- A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspective (NeurIPS 2025 · similarity 0.83)
- Learning With Multi-Group Guarantees For Clusterable Subpopulations (ICML 2025 · similarity 0.83)
- When Models Don’t Collapse: On the Consistency of Iterative MLE (NeurIPS 2025 · similarity 0.83)
- Unified Scaling Laws for Compressed Representations (NeurIPS 2025 · 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.
