PAC-Bayesian Spectrally-Normalized Bounds for Adversarially Robust Generalization.
Jiancong XiaoRuoyu SunZhi-Quan LuoPublished in: CoRR (2023)
Keyphrases
- pac bayesian
- generalization bounds
- distribution free
- data dependent
- learning theory
- generalization ability
- rademacher complexity
- upper bound
- error bounds
- statistical learning theory
- model selection
- machine learning
- large deviations
- linear classifiers
- parameter free
- normal distribution
- random variables
- similarity measure
- decision trees