Uniform Convergence of Interpolators: Gaussian Width, Norm Bounds, and Benign Overfitting.
Frederic KoehlerLijia ZhouDanica J. SutherlandNathan SrebroPublished in: CoRR (2021)
Keyphrases
- uniform convergence
- generalization error
- covering numbers
- empirical risk minimization
- gaussian kernels
- sufficient conditions
- generalization bounds
- vc dimension
- reproducing kernel hilbert space
- learning rate
- cross validation
- upper and lower bounds
- upper bound
- risk minimization
- real valued
- gaussian kernel
- statistical learning theory
- maximum likelihood
- sample complexity
- active learning
- large deviations
- learning algorithm
- lower bound
- data dependent
- objective function
- training data
- decision trees
- special case
- linear classifiers
- training set
- inductive inference
- gaussian process
- generalization ability
- distribution free
- learning theory
- ranking algorithm
- loss function