preTrainDArch: Pre-trains a 'DArch' network

In darch: Package for Deep Architectures and Restricted Boltzmann Machines

Description

This function pre-trains a DArch network with the contrastive divergence method

Usage

preTrainDArch(darch, dataSet, dataSetValid = NULL, numEpochs = 1,
  numCD = 1, lastLayer = 0, isClass = F, consecutive = T, ...)

Arguments

darch	A instance of the class DArch.
dataSet	DataSet to be used for training.
dataSetValid	DataSet to be used for validation.
numEpochs	The number of epochs
numCD	The number of CD iterations
lastLayer	Numeric indicating after which layer to stop training.
isClass	Whether to test pre-trained networks against target data
consecutive	Whether to train RBMs consecutively instead of each one epoch at a time.
...	Additional parameters for the function trainRBM

Details

The function runs for every RBM in the attribute rbmList the training function trainRBM copies after the training the weights and biases into the corresponding layer of the DArch network.

Value

Trained DArch instance

See Also

DArch, RBM, trainRBM

darch documentation built on May 29, 2017, 8:14 p.m.