C# Class MyMediaLite.Util.NelderMead

Nealder-Mead algorithm for finding suitable hyperparameters

Exibir arquivo Open project: zenogantner/MML-KDD

Public Methods

Method	Description
FindMinimum ( string evaluation_measure, IList hp_names, IList initial_hp_values, RatingPredictor recommender, ISplit split ) : double	Find the the parameters resulting in the minimal results for a given evaluation measure The recommender will be set to the best parameter value after calling this method.
FindMinimum ( string error_measure, RatingPredictor recommender ) : double	Find best hyperparameter (according to an error measure) using Nelder-Mead search

Method	Description
ComputeCenter ( double>.Dictionary results, Vector>.Dictionary hp_values ) : Vector
CreateConfigString ( IList hp_names, IList hp_values ) : string
Run ( RatingPredictor recommender, ISplit split, string hp_string, string evaluation_measure ) : double

Find the the parameters resulting in the minimal results for a given evaluation measure

The recommender will be set to the best parameter value after calling this method.

public static FindMinimum ( string evaluation_measure, IList hp_names, IList initial_hp_values, RatingPredictor recommender, ISplit split ) : double
evaluation_measure	string	the name of the evaluation measure
hp_names	IList	the names of the hyperparameters to optimize
initial_hp_values	IList	the values of the hyperparameters to try out first
recommender	MyMediaLite.RatingPrediction.RatingPredictor	the recommender
split	ISplit	the dataset split to use
return	double

Find best hyperparameter (according to an error measure) using Nelder-Mead search

public static FindMinimum ( string error_measure, RatingPredictor recommender ) : double
error_measure	string	an error measure (lower is better)
recommender	MyMediaLite.RatingPrediction.RatingPredictor	a rating predictor (will be set to best hyperparameter combination)
return	double