SAENET.predict: Obtain the compressed representation of new data for...

In SAENET: A Stacked Autoencoder Implementation with Interface to 'neuralnet'

Description

Obtain the compressed representation of new data for specified layers from a stacked autoencoder.

Usage

SAENET.predict(h, new.data, layers = c(1), all.layers = FALSE)

Arguments

h	The object returned from SAENET.train()
new.data	A matrix of training data.
layers	A numeric vector indicating which layers of the stacked autoencoder to return output for
all.layers	A boolean value indicating whether to override layers and return the encoded output for all layers. Defaults to FALSE

Value

A list, for which each element corresponds to the output of predict.autoencoder() from package autoencoder for the specified layers of the stacked autoencoder.

Examples

library(autoencoder)
data(iris)
#### Train a stacked sparse autoencoder with a (5,3) architecture and
#### a relatively minor sparsity penalty. Try experimenting with the
#### lambda and beta parameters if you haven't worked with sparse
#### autoencoders before - it's worth inspecting the final layer
#### to ensure that output activations haven't simply converged to the value of
#### rho that you gave (which is the desired activation level on average).
#### If the lambda/beta parameters are set high, this is likely to happen.


output <- SAENET.train(as.matrix(iris[1:100,1:4]), n.nodes = c(5,3),
                       lambda = 1e-5, beta = 1e-5, rho = 0.01, epsilon = 0.01)


predict.out <- SAENET.predict(output, as.matrix(iris[101:150,1:4]), layers = c(2))

Example output

autoencoding...
Optimizer counts:
function gradient 
      19       17 
Optimizer: successful convergence.
Optimizer: convergence = 0, message = 
J.init = 20.07615, J.final = 15.44573, mean(rho.hat.final) = 0.9998588
autoencoding...
Optimizer counts:
function gradient 
     191      187 
Optimizer: successful convergence.
Optimizer: convergence = 0, message = 
J.init = 0.6235624, J.final = 7.775034e-08, mean(rho.hat.final) = 0.01018348

SAENET documentation built on May 30, 2017, 1:51 a.m.