The Most Robust Loss Function for Boosting.
Takafumi KanamoriTakashi TakenouchiShinto EguchiNoboru MurataPublished in: ICONIP (2004)
Keyphrases
- loss function
- boosting framework
- boosting algorithms
- gradient boosting
- pairwise
- logistic regression
- support vector
- learning to rank
- risk minimization
- regularization term
- empirical risk
- convex loss functions
- solution path
- hinge loss
- reproducing kernel hilbert space
- pairwise constraints
- base learners
- distance measure
- special case
- machine learning