A Survey on Data-Dependent Worst-Case Generalization Bounds
数据依赖性最坏情况泛化界限的综述
机构 * École polytechnique(巴黎政治学院)
AI总结 本文综述了通过扩展PAC-Bayesian理论、优化轨迹的几何拓扑描述以及稳定性假设来改进数据依赖性最坏情况泛化界限的方法。
Comments 15 pages, 4 figures, 3 tables. The LaTeX source uses the JMLR preprint style (jmlr2e.sty) and BibTeX (refs.bib). Central references in arXiv form include arXiv:2404.17442, arXiv:2006.09313, arXiv:2302.02766, arXiv:2407.08723, and arXiv:2507.06775