StackRBM: Stacked Restricted Boltzmann Machine
In TimoMatzen/RBM: Package for fitting RBM and DBN models in R.

Description Usage Arguments Value Examples

Function to stack several Restricted Boltzmann Machines, trained greedily by training a RBM (using the RBM function) at each layer and then using the output of that RBM to train the next layer RBM.

1
2
3

StackRBM(x, y, n.iter = 100, layers = c(100, 100, 30),
  learning.rate = 0.1, size.minibatch = 10, lambda = 0.1,
  momentum = 0.5)

`x`	A matrix with binary features of shape samples * features.
`y`	A matrix with labels for the data, only when the last layer is a classification RBM. (Optional)
`n.iter`	Number of epochs for training each RBM layer.
`layers`	Vector with the number of hidden nodes for each RBM layer.
`learning.rate`	The learning rate for training each RBM layer.
`size.minibatch`	The size of the minibatches used for training.
`lambda`	The sparsity penalty lambda to prevent the system from overfitting.
`momentum`	Speeds up the gradient descent learning.

A list with the trained weights of the stacked RBM that can be used for the predict RBM function when a classification RBM is at the top layer of the ReconstructRBM function to reconstruct data with the model.

# Load MNIST data
data(MNIST)

# Train a unsupervised Stack of 3 RBMs
mod <- StackRBM(MNIST$trainX, layers = c(100,100,100))

# Classification RBM as top layer
modSup <- StackRBM(MNIST$trainX, MNIST$trainY, layers = c(100,100,100))