|
|||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | ||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |
java.lang.Objectorg.mymedialite.hyperparameter.NelderMead
public class NelderMead
Nealder-Mead algorithm for finding suitable hyperparameters.
Method Summary | |
---|---|
static double |
findMinimum(java.lang.String evaluation_measure,
java.util.List<java.lang.String> hp_names,
java.util.List<DoubleMatrix1D> initial_hp_values,
RatingPredictor recommender,
ISplit<IRatings> split)
Find the the parameters resulting in the minimal results for a given evaluation measure. |
static double |
findMinimum(java.lang.String error_measure,
RatingPredictor recommender)
Find best hyperparameter (according to an error measure) using Nelder-Mead search. |
Methods inherited from class java.lang.Object |
---|
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
Method Detail |
---|
public static double findMinimum(java.lang.String error_measure, RatingPredictor recommender) throws java.lang.Exception
error_measure
- an error measure (lower is better)recommender
- a rating predictor (will be set to best hyperparameter combination)
java.lang.Exception
public static double findMinimum(java.lang.String evaluation_measure, java.util.List<java.lang.String> hp_names, java.util.List<DoubleMatrix1D> initial_hp_values, RatingPredictor recommender, ISplit<IRatings> split) throws java.lang.Exception
evaluation_measure
- the name of the evaluation measurehp_names
- the names of the hyperparameters to optimizeinitial_hp_values
- the values of the hyperparameters to try out firstrecommender
- the recommendersplit
- the dataset split to use
java.lang.Exception
|
|||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | ||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |