libsvm

Learning

training set file
testing set file
validation set file
output of the trained model file
Yes No self-consistency evaluation over training set
seed single random number generator
Yes No strict checking of dataset aganist algorithm

LIBSVM

set type of SVM
set type of kernel function
set degree in polynomial kernel
set gamma in gaussian kernel
set gamma in polynomial kernel
set gamma in sigmoid kernel
set coef0 in polynomial kernel
set coef0 in sigmoid kernel
set the cost parameter for linear kernel
set the cost parameter for gaussian kernel
set the cost parameter for polynomial kernel
set the cost parameter for sigmoid kernel
set the parameter nu of nu-SVC, one-class SVM, and nu-SVR
set the epsilon in loss function of epsilon-SVR
set cache memory size in MB
set tolerance of termination criterion
Yes No whether to use the shrinking heuristics
set positive weight for linear kernel
set positive weight for gaussian kernel
set positive weight for sigmoid kernel
set positive weight for polynomial kernel
set negative weight for linear kernel
set negative weight for gaussian kernel
set negative weight for sigmoid kernel
set negative weight for polynomial kernel

Selection

number of folds in n-fold cross-validation
the number of times to repeat the validation
the type of validation
Yes No use stratifed validation
Yes No force instance level validation for bagged set
Yes No force instances in positive bag to be positive
increment for progressive validation
validation model file
metric used to select best parameters
Yes No evaluate on the instance level
what to print while tuning

Validation

number of folds in n-fold cross-validation
the number of times to repeat the validation
the type of validation
Yes No use stratifed validation
Yes No force instance level validation for bagged set
Yes No force instances in positive bag to be positive
increment for progressive validation
validation model file

Format

index of class in feature vector
number of labels prefixing the feature
Dataset has header

Transform

Yes No use the product to expand the attribute space
list of classes to set positive, all else negative
name of positive class
name of negative class
begining of bag substring
end of bag substring