Multi-Class Classification Using a New Sigmoid Loss Function for Minimum Classification Error (MCE).
Madhavi Vedula RatnagiriLawrence R. RabinerBiing-Hwang JuangPublished in: ICMLA (2010)
Keyphrases
- minimum classification error
- loss function
- multi class classification
- support vector
- statistical pattern recognition
- multi class
- pairwise
- classification error
- binary classification
- maximum likelihood estimation
- hidden markov models
- discriminative training
- belief networks
- multi class boosting
- gaussian mixture
- multi task learning
- cross validation
- hyperparameters
- kernel methods
- classification accuracy
- bayesian networks
- learning problems
- data sets
- training examples
- text categorization
- em algorithm
- denoising
- support vector machine