Research map / Optimization / Implicit bias and generalization
Instance-Dependent Generalization Bounds via Optimal Transport
Songyan Hou, Parnian Kassraie, Anastasis Kratsios, Andreas Krause, Jonas Rothfuss
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 a partitioned local-Lipschitz bound instead of the global Lipschitz bound | partition bound · classification error bound, N=2560 | generalization bound: 18.407 -> 1.447 | Improved |
| use a partitioned local-Lipschitz bound instead of the global Lipschitz bound | partition bound · classification error bound, N=81920 | generalization bound: 2.856 -> 0.478 | Improved |
| use partition bound instead of the uniform Rademacher bound | partition bound · classification error bound, N=2560 | generalization bound: 39.545 -> 1.447 | Improved |
| use partition bound instead of the uniform Rademacher bound | partition bound · classification error bound, N=81920 | generalization bound: 16.738 -> 0.478 | Improved |
| increase the number of training points from 2560 to 81920 | partition bound · classification error bound | generalization bound: 1.447 -> 0.478 | Improved |
| partition the input domain into parts finer and finer (ablation) | partition bound · regression and classification | generalization bound: bound values initially decrease and, as the partitioning becomes much finer, increase again | Mixed |
| increase network size by a factor 1000 in parameters (ablation) | partition bound · regression and classification | generalization bound: only a minor effect on the value of the bound | No clear effect |
Closest papers by meaning
- Fantastic Generalization Measures are Nowhere to be Found (ICLR 2024 · similarity 0.91)
- Generalizability of Neural Networks Minimizing Empirical Risk Based on Expressive Power (ICLR 2025 · similarity 0.90)
- Generalization Bound of Gradient Flow through Training Trajectory and Data-dependent Kernel (NeurIPS 2025 · similarity 0.89)
- Slicing Mutual Information Generalization Bounds for Neural Networks (ICML 2024 · similarity 0.89)
- Generalization Bounds for Rank-sparse Neural Networks (NeurIPS 2025 · similarity 0.88)
- 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.
