The Perils of Being Unhinged: On the Accuracy of Classifiers Minimizing a Noise-Robust Convex Loss.
Philip M. LongRocco A. ServedioPublished in: Neural Comput. (2022)
Keyphrases
- image noise
- highly accurate
- registration errors
- training data
- training and testing data
- high accuracy
- highest accuracy
- individual classifiers
- computational cost
- convex functions
- noise immunity
- confusion matrix
- higher classification accuracy
- classification accuracy
- noisy environments
- global optimality
- geometric distortions
- fold cross validation
- salt pepper
- gaussian noise
- estimation error
- convex optimization
- noisy data
- error rate
- convex sets
- linear classifiers
- roc curve
- watermarking scheme
- noise level
- noise reduction
- detection rate
- ensemble pruning
- convex combinations
- prediction accuracy
- decision trees