Net-class

This is a abstract class for neural networks. It provides some
functionalities used in more than one network type.

The darch package is built on the basis of the code from G. E.
Hinton and R. R. Salakhutdinov (available under Matlab Code for deep belief
nets). This package is for generating neural networks with many layers (deep
architectures) and train them with the method introduced by the publications
"A fast learning algorithm for deep belief nets" (G. E. Hinton, S. Osindero,
Y. W. Teh; doi: 10.1162/neco.2006.18.7.1527) and "Reducing the
dimensionality of data with neural networks" (G. E. Hinton, R. R.
Salakhutdinov; doi: 10.1126/science.1127647). This method includes a pre
training with the contrastive divergence method published by G.E Hinton
(2002; doi: 10.1162/089976602760128018) and a fine tuning with common known
training algorithms like backpropagation or conjugate gradients.
Additionally, supervised fine-tuning can be enhanced with maxout and
dropout, two recently developed techniques to improve fine-tuning for deep
learning.

Martin Drees

darch

Package for Deep Architectures and Restricted Boltzmann Machines

Net-class function

<dl class="dl-horizontal">
<dt><code>batchSize</code></dt><dd>Object of class <code>"numeric"</code>. The batch size
for the training and test data during the learning.</dd></dl><dt><code>errorFunction</code></dt><dd>Object of class <code>"function"</code>. Function for error
calculation.</dd><dt><code>ff</code></dt><dd>Object of class <code>"logical"</code>. Indicates if the package
<code><a rd-options="ff" href="/link/ff?package=darch&version=0.10.0&to=ff" data-mini-rdoc="ff::ff">ff</a></code> is used to save the network data.</dd><dt><code>genWeightFunction</code></dt><dd>Object of class <code>"function"</code>. A function to
generate a randomly initialised weight matrix.</dd><dt><code>normalizeWeights</code></dt><dd>Logical indicating whether to normalize the weights.</dd><dt><code>learnRateWeights</code></dt><dd>Learn rate.</dd><dt><code>initialMomentum</code></dt><dd>Initial momentum.</dd><dt><code>finalMomentum</code></dt><dd>Final momentum.</dd><dt><code>momentumSwitch</code></dt><dd>Epoch at which to switch from <code>initialMomentum</code> to
<code>finalMomentum</code>.</dd><dt><code>epochs</code></dt><dd>Number of epochs.</dd><dt><code>stats</code></dt><dd>Training statistics.</dd>

Net-class: Abstract class for neural networks.

Description

Arguments

Slots

See Also