Classification vs regression in overparameterized regimes: Does the loss function matter?
Vidya MuthukumarAdhyyan NarangVignesh SubramanianMikhail BelkinDaniel J. HsuAnant SahaiPublished in: CoRR (2020)
Keyphrases
- loss function
- support vector
- hinge loss
- reproducing kernel hilbert space
- gradient boosting
- pairwise
- feature vectors
- risk minimization
- regression problems
- benchmark datasets
- model selection
- support vector machine
- classification accuracy
- boosting algorithms
- svm classifier
- decision trees
- machine learning
- bayes rule
- support vector machine svm
- base learners
- feature space
- cost sensitive
- empirical risk
- boosting framework
- logistic regression
- text classification
- convex loss functions
- pattern classification
- decision function
- pairwise constraints
- generalization ability
- linear regression
- learning to rank
- cross validation
- regression model
- feature selection