Page 259 - PVT Property Correlations

228 PVT Property Correlations

FIGURE 10.4 Two-
layer ANN. ANN, arti-
ficial neural network.

FIGURE 10.5 Multilayer deep ANN. ANN, artificial neural network.

The same basic calculations are performed for any type or structure of
ANN. Figs. 10.3 10.5 represent three different ANN structures. The first is an
example of an ANN with one hidden layer; the second represents a network
with two hidden layers, while the third represents a deep network with multiple
hidden layers. The input nodes are multiple in each of the networks. The output
nodes can be one or more. In general, additional hidden layers are needed to
decipher complex relations between inputs and output(s). However, additional
hidden layers require also that large number of training records be available.
The earlier ANN example models show differences not only in the num-
ber of hidden layers, but also in node communication. The first ANN is
characterized by the concept of “full communication,” whereby all nodes
in the input layer communicate with all nodes in the hidden layer, and
all nodes in the hidden layer communicate with all nodes in the output
layer. The second and third ANN models are examples of partial communi-
cation networks.