A Conjugate Property between Loss Functions and Uncertainty Sets in Classification Problems
Takafumi KanamoriAkiko TakedaTaiji SuzukiPublished in: CoRR (2012)
Keyphrases
- loss function
- pairwise
- loss minimization
- squared error
- support vector
- logistic regression
- learning to rank
- reproducing kernel hilbert space
- risk minimization
- boosting framework
- empirical risk
- learning models
- convex loss functions
- hinge loss
- boosting algorithms
- conditional probabilities
- machine learning
- data dependent
- update rules
- information retrieval