Improving Bayesian credibility intervals for classifier error rates using maximum entropy empirical priors.
Mats G. GustafssonMikael WallmanUlrika Wickenberg-BolinHanna GöranssonMårten FryknäsClaes R. AnderssonAnders IsakssonPublished in: Artif. Intell. Medicine (2010)
Keyphrases
- maximum entropy
- class conditional
- markov models
- maximum entropy principle
- expected error
- maximum entropy model
- principle of maximum entropy
- iterative scaling
- prior probabilities
- transformation based learning
- feature selection
- training data
- named entity recognition
- conditional random fields
- training set
- machine learning
- minimum cross entropy
- learning algorithm
- class labels
- markov random field
- data clustering
- information theoretic
- class probabilities
- training samples
- maximum likelihood
- support vector machine
- pairwise
- feature space
- decision trees