A methodology for training set instance selection using mutual information in time series prediction.
Milos B. StojanovicMilos BozicMilena M. StankovicZoran StajicPublished in: Neurocomputing (2014)
Keyphrases
- instance selection
- mutual information
- training set
- nearest neighbor
- classification accuracy
- supervised learning
- feature selection
- feature and instance selection
- image registration
- multiple instance learning
- test set
- data reduction
- selection criterion
- data sets
- artificial neural networks
- feature space
- training data
- similarity measure
- non stationary
- knowledge discovery and data mining
- text classification
- semi supervised learning
- multi class
- active learning
- unlabeled data
- training samples
- training examples
- support vector machine
- machine learning
- unsupervised learning
- cross validation
- multi modal
- classification algorithm
- concept drift
- learning algorithm
- neural network