do ANN model training

training.ANN {machineLearning}

R Documentation

do ANN model training

Description

Usage

training.ANN(trainSet,
    hiddenSize = [100,30],
    learnRate = 0.1,
    momentum = 0.9,
    weight0 = "random",
    active = MLkit.activation,
    normalMethod = RelativeScaler,
    learnRateDecay = 1E-10,
    truncate = -1,
    softmax = TRUE,
    selectiveMode = FALSE,
    dropout = 0,
    maxIterations = 10000,
    minErr = 0.01,
    parallel = TRUE,
    outputSnapshot = FALSE,
    attribute = -1);

Arguments

trainSet

A dataset object that used for ANN model training. [as DataSet]

hiddenSize

An integer vector for indicates the network size of the hidden layers in the ANN network.

learnRate

[as double]

momentum

[as double]

weight0

weight method for initialize the ANN network model.

active

[as activation]

normalMethod

[as Methods]

learnRateDecay

[as double]

truncate

[as double]

selectiveMode

[as boolean]

maxIterations

[as integer]

minErr

[as double]

parallel

[as boolean]

outputSnapshot

this parameter will config the output object type. this function is returns the raw ANN model by default, and you can change the output type to file model by set this parameter value to TRUE. [as boolean]

attribute

run training for a single output or all of the result output. [as integer]

Authors

MLkit

Value

this function returns data object in these one of the listed data types: NeuralNetwork, Network.

clr value class

[Package machineLearning version 1.0.0.0 Index]