Research map / Optimization / Implicit bias and generalization
Generalization Bounds for Noisy Iterative Algorithms Using Properties of Additive Noise Channels
Hao Wang, Rui Gao, Flavio P. Calmon
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 the variance-based generalization bound instead of the gradient-prediction-residual bound | generalization bound · MNIST | Kendall's tau: 0.25 -> 0.50 | Improved |
| Use the variance-based generalization bound instead of the gradient-prediction-residual bound | generalization bound · CIFAR-10 | Kendall's tau: 0.38 -> 0.45 | Improved |
| Use the variance-based bound instead of the baseline bound, conditional independence test | generalization bound · MNIST | MI: 0.12 -> 0.34 | Improved |
| Use the variance-based bound instead of the baseline bound, varying network width | generalization bound · MNIST | correlation, width varied: 0.26 -> 1.00 | Improved |
| Increase label corruption level from 0% to 75% of training samples | SGLD-trained network · corrupted labels | generalization bound and gap: bound and gap both increase with corruption level | Hurt |
| Increase network width | SGLD-trained network · network width experiment | generalization gap and bound: both decrease with width | Improved |
Closest papers by meaning
- On Generalization Error Bounds of Noisy Gradient Methods for Non-Convex Learning (ICLR 2020 · similarity 0.91)
- Analyzing the Generalization Capability of SGLD Using Properties of Gaussian Channels (NeurIPS 2021 · similarity 0.91)
- Sharpened Generalization Bounds based on Conditional Mutual Information and an Application to Noisy, Iterative Algorithms (NeurIPS 2020 · similarity 0.91)
- Exactly Tight Information-theoretic Generalization Bounds via Binary Jensen-Shannon Divergence (ICML 2025 · similarity 0.90)
- Generalization of noisy SGD in unbounded non-convex settings (ICML 2025 · similarity 0.90)
- Towards Generalization beyond Pointwise Learning: A Unified Information-theoretic Perspective (ICML 2024 · similarity 0.88)
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.
