A Novel Wrapper Approach for Feature Selection in Object-Based Image Classification Using Polygon-Based Cross-Validation.
Lei MaManchun LiYu GaoTan ChenXiaoxue MaLean QuPublished in: IEEE Geosci. Remote. Sens. Lett. (2017)
Keyphrases
- cross validation
- feature selection
- image classification
- model selection
- support vector
- feature extraction
- variable selection
- hyperparameters
- mutual information
- text categorization
- training set
- image features
- machine learning
- generalization error
- classification accuracy
- feature space
- selected features
- feature ranking
- cross validated
- text classification
- regression problems
- regularization parameter
- sample size
- multi label
- meta learning
- feature selection algorithms
- leave one out cross validation
- support vector machine
- wrapper feature selection
- wrapper method
- information criterion
- feature set
- svm classifier
- error estimates
- unseen data
- multi class
- dimensionality reduction
- naive bayes
- kernel function
- gaussian process
- knn
- nearest neighbor classifiers
- feature subset
- small sample
- svm classification
- ls svm
- logistic regression