Package milk.classifiers

Interface Summary
MITransform Interface to transform any MI data into data suitable for mono-instance learner
MIUpdateableClassifier Interface to incremental classification models that can learn using one exemplar at a time.
 

Class Summary
DD Re-implement the Diverse Density algorithm, changes the testing procedure.
MDD Modified DD, with collective assumption
MIBoost MI AdaBoost method, consider the geometric mean of posterior of instances inside a bag (arithmatic mean of log-posterior) and the expectation for a bag is taken inside the loss function.
MIClassifier Abstract classifier.
MIEvaluation Class for evaluating machine learning models.
MILR Use MI assumption but within LR Valid options are:
MILRARITH Using collective assumption, arithmatic average of the posteriors of instances inside a bag is taken.
MILRGEOM Normalized goemetric mean is taken on the posteriors of instances, regardless of class labels Valid options are:
MINND 0657.591B Dissertation Multiple-Instance Nearest Neighbour with Distribution learner .
MIRBFNetwork Multi-instance RBF network.
MIWrapper Weighted Wrapper method from Eibe Valid options are:
SimpleMI Avg of feature data and reduce MI data into mono-instance Valid options are:
TLD 0657.594 Thesis Two-Level Distribution approach, changes the starting value of the searching algorithm, supplement the cut-off modification and check missing values.
TLDSimple 0657.594 Thesis A simpler version of TLD, \mu random but \sigma^2 fixed and estimated via data