|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |
java.lang.Objectorg.apache.mahout.classifier.discriminative.LinearTrainer
public abstract class LinearTrainer
Implementors of this class need to provide a way to train linear discriminative classifiers. As this is just the reference implementation we assume that the dataset fits into main memory - this should be the first thing to change when switching to Hadoop.
Constructor Summary | |
---|---|
protected |
LinearTrainer(int dimension,
double threshold,
double init,
double initBias)
Initialize the trainer. |
Method Summary | |
---|---|
LinearModel |
getModel()
Retrieves the trained model if called after train, otherwise the raw model. |
void |
train(Vector labelset,
Matrix dataset)
Initializes training. |
protected abstract void |
update(double label,
Vector dataPoint,
LinearModel model)
Implement this method to match your training strategy. |
Methods inherited from class java.lang.Object |
---|
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
Constructor Detail |
---|
protected LinearTrainer(int dimension, double threshold, double init, double initBias) throws CardinalityException
dimension
- number of expected features.threshold
- threshold to use for classification.init
- initial value of weight vector.initBias
- initial classification bias.
CardinalityException
Method Detail |
---|
public void train(Vector labelset, Matrix dataset) throws IndexException, CardinalityException, TrainingException
dataset
- the dataset to train on. Each column is treated as point.labelset
- the set of labels, one for each data point. If the cardinalities
of data- and labelset do not match, a CardinalityException is
thrown
IndexException
CardinalityException
TrainingException
public LinearModel getModel()
protected abstract void update(double label, Vector dataPoint, LinearModel model)
model
- the model to update.label
- the target label of the wrongly classified data point.dataPoint
- the data point that was classified incorrectly.
|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |