Adjusting the Outputs of a Classifier to New a Priori Probabilities May Significantly Improve Classification Accuracy: Evidence from a multi-class problem in remote sensing.
Patrice LatinneMarco SaerensChristine DecaesteckerPublished in: ICML (2001)
Keyphrases
- remote sensing
- multi class
- class probabilities
- classification accuracy
- feature selection
- multi class classifier
- binary classifiers
- multiclass classification
- support vector machine
- linear classifiers
- error correcting output codes
- multispectral
- remote sensing images
- change detection
- cost sensitive
- multi class svm
- training data
- support vector
- feature space
- image analysis
- binary classification
- probability estimates
- automatic image registration
- feature set
- binary classification problems
- training set
- image fusion
- hyperspectral
- land cover
- high resolution
- satellite data
- remote sensing data
- satellite images
- image processing
- naive bayes
- multi class classification
- single class
- earth observation
- multiple classes
- feature subset
- pairwise
- object detection
- svm classifier
- base classifiers
- multi class problems
- multi label classification
- data sets
- pattern recognition
- multiscale
- unsupervised learning
- training samples
- decision trees
- text categorization
- kernel function