predict.stacked_rbm: Predict from a Stacked Restricted Boltzmann Machine

In zachmayer/rbm: Restricted Boltzmann Machines

Description

This function takes a stacked RBM and a matrix of new data, and predicts for the new data with the RBM.

Usage

## S3 method for class 'stacked_rbm'
predict(object, newdata, type = "probs",
  omit_bias = TRUE, ...)

Arguments

object	a RBM object
newdata	a sparse matrix of new data
type	a character vector specifying whether to return the hidden unit activations, hidden unit probs, or hidden unit states. Activations or probabilities are typically the most useful if you wish to use the RBM features as input to another predictive model (or another RBM!). Note that the hidden states are stochastic, and may be different each time you run the predict function, unless you set random.seed() before making predictions. Activations and states are non-stochastic, and will be the same each time you run predict.
omit_bias	Don't return the bias column in the prediciton matrix.
...	not used

Value

a sparse matrix

zachmayer/rbm documentation built on May 4, 2019, 9:07 p.m.