Minimum classification error vs. maximum margin: How should we penalize unseen samples?
Shigeru KatagiriHideyuki WatanabePublished in: CIP (2012)
Keyphrases
- maximum margin
- minimum classification error
- training set
- hyperplane
- support vector
- structured prediction
- learning algorithm
- classification error
- markov networks
- training samples
- support vector machine
- training examples
- generalization error
- pattern classification
- maximum likelihood
- data sets
- multi task
- prior knowledge
- feature selection
- decision trees
- gaussian mixture
- class labels
- supervised learning
- small number