dbn.train: Training a Deep Belief Net

Description Usage Arguments Author(s)

View source: R/dbn_train.R

Description

Training a Deep Belief Net

Usage

1
2
3
dbn.train(x, hidden = c(10, 10), numepochs = 3, batchsize = 100,
  learningrate = 0.8, learningrate_scale = 1, momentum = 0.5,
  visible_type = "bin", hidden_type = "bin", cd = 1)

Arguments

x

matrix of x values for examples

hidden

number of hidden units

numepochs

number of iteration for samples Default is 3.

batchsize

size of mini-batch. Default is 100.

learningrate

learning rate for gradient descent. Default is 0.8.

learningrate_scale

learning rate will be mutiplied by this scale after every iteration. Default is 1 .

momentum

momentum for gradient descent. Default is 0.5 .

visible_type

activation function of input unit.Only support "sigm" now

hidden_type

activation function of hidden unit.Only support "sigm" now

cd

number of iteration for Gibbs sample of CD algorithm.

Author(s)

Xiao Rong


DimitriF/DLC documentation built on Oct. 7, 2017, 11:10 p.m.