Bagging Provides Assumption-free Stability
This provides a theoretical foundation for bagging's stability, benefiting practitioners using ensemble methods in machine learning, though it is incremental as it builds on existing bagging concepts.
The paper derived a finite-sample guarantee on the stability of bagging for any model without assumptions on data distribution, base algorithm properties, or covariate dimensionality, showing it is optimal up to a constant and empirically stabilizes even highly unstable algorithms.
Bagging is an important technique for stabilizing machine learning models. In this paper, we derive a finite-sample guarantee on the stability of bagging for any model. Our result places no assumptions on the distribution of the data, on the properties of the base algorithm, or on the dimensionality of the covariates. Our guarantee applies to many variants of bagging and is optimal up to a constant. Empirical results validate our findings, showing that bagging successfully stabilizes even highly unstable base algorithms.