Package com.googlecode.fannj

Class Trainer

  • public class Trainer
extends java.lang.Object
    Trains an ANN. Currently only File based training is supported.

    • Constructor Summary

      Constructors 
      Constructor Description
      Trainer​(Fann fann)  

    • Method Summary

      All Methods Static Methods Instance Methods Concrete Methods 
      Modifier and Type Method Description
      float cascadeTrain​(java.lang.String dataFile, int maxNeurons, int neuronsBetweenReports, float desiredError)  
      protected static void fann_cascadetrain_on_file​(com.sun.jna.Pointer ann, java.lang.String filename, int max_neurons, int neurons_between_reports, float desired_error)  
      protected static void fann_destroy_train​(com.sun.jna.Pointer data)
      Deallocate the data
      protected static float fann_get_MSE​(com.sun.jna.Pointer ann)
      Reads the mean square error from the network.
      protected static int fann_get_training_algorithm​(com.sun.jna.Pointer ann)  
      protected static com.sun.jna.Pointer fann_read_train_from_file​(java.lang.String filename)
      Read the training or testing data from a file You must call fann_destroy_train(Pointer) on the Pointer you get from this after you have finished with it
      protected static void fann_reset_MSE​(com.sun.jna.Pointer ann)
      Resets the mean square error from the network.
      protected static void fann_set_training_algorithm​(com.sun.jna.Pointer ann, int training_algorithm)  
      protected static float fann_test_data​(com.sun.jna.Pointer ann, com.sun.jna.Pointer data)
      Test the network using data and return the MSE of the network.
      protected static void fann_train_on_file​(com.sun.jna.Pointer ann, java.lang.String filename, int max_epochs, int epochs_between_reports, float desired_error)  
      void setTrainingAlgorithm​(TrainingAlgorithm algorithm)  
      float test​(java.lang.String testingFile)  
      float train​(java.lang.String trainingFile, int maxEpochs, int epochsBetweenReports, float desiredError)  

      • Methods inherited from class java.lang.Object

        clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

    • Constructor Detail

      • Trainer

        public Trainer​(Fann fann)

    • Method Detail

      • train

        public float train​(java.lang.String trainingFile,
                   int maxEpochs,
                   int epochsBetweenReports,
                   float desiredError)
        Parameters:
        trainingFile -
        maxEpochs -
        epochsBetweenReports -
        desiredError -
        Returns:
        MSE for the ann once trained

      • cascadeTrain

        public float cascadeTrain​(java.lang.String dataFile,
                          int maxNeurons,
                          int neuronsBetweenReports,
                          float desiredError)
        Parameters:
        dataFile -
        maxNeurons -
        neuronsBetweenReports -
        desiredError -
        Returns:
        MSE for the ann once trained

      • setTrainingAlgorithm

        public void setTrainingAlgorithm​(TrainingAlgorithm algorithm)

      • test

        public float test​(java.lang.String testingFile)
        Parameters:
        testingFile -
        Returns:
        MSE for the Fann which has been tested

      • fann_train_on_file

        protected static void fann_train_on_file​(com.sun.jna.Pointer ann,
                                         java.lang.String filename,
                                         int max_epochs,
                                         int epochs_between_reports,
                                         float desired_error)

      • fann_cascadetrain_on_file

        protected static void fann_cascadetrain_on_file​(com.sun.jna.Pointer ann,
                                                java.lang.String filename,
                                                int max_neurons,
                                                int neurons_between_reports,
                                                float desired_error)

      • fann_set_training_algorithm

        protected static void fann_set_training_algorithm​(com.sun.jna.Pointer ann,
                                                  int training_algorithm)

      • fann_get_training_algorithm

        protected static int fann_get_training_algorithm​(com.sun.jna.Pointer ann)

      • fann_reset_MSE

        protected static void fann_reset_MSE​(com.sun.jna.Pointer ann)
        Resets the mean square error from the network.
        Parameters:
        ann -

      • fann_get_MSE

        protected static float fann_get_MSE​(com.sun.jna.Pointer ann)
        Reads the mean square error from the network.
        Parameters:
        ann -
        Returns:
        the mean square error of the network

      • fann_test_data

        protected static float fann_test_data​(com.sun.jna.Pointer ann,
                                      com.sun.jna.Pointer data)
        Test the network using data and return the MSE of the network. You might need to run fann_reset_MSE(Pointer) first
        Parameters:
        ann -
        data - the data to test with
        Returns:
        the mean square error of the network

      • fann_read_train_from_file

        protected static com.sun.jna.Pointer fann_read_train_from_file​(java.lang.String filename)
        Read the training or testing data from a file You must call fann_destroy_train(Pointer) on the Pointer you get from this after you have finished with it
        Parameters:
        filename - the file name of the file to read the data from
        Returns:
        pointer to the data which has been read for use with fann_test_data(Pointer,Pointer)

      • fann_destroy_train

        protected static void fann_destroy_train​(com.sun.jna.Pointer data)
        Deallocate the data
        Parameters:
        data - the training/testing data to deallocate