A Constructive Prediction of the Generalization Error Across Scales.
Jonathan S. RosenfeldAmir RosenfeldYonatan BelinkovNir ShavitPublished in: ICLR (2020)
Keyphrases
- generalization error
- cross validation
- model selection
- learning algorithm
- prediction accuracy
- upper bound
- sample complexity
- binary classification
- sample size
- classification error
- training data
- active learning
- linear classifiers
- training error
- generalization error bounds
- training set
- data sets
- supervised learning
- target function
- conditional expectation
- subspace information criterion
- learning machines
- prediction error
- perceptron learning
- boosting algorithms
- training set size
- low error
- multi class
- face recognition
- machine learning