8.1 Feed-Forward Networks

Figure 8.2 shows a feed-forward network with one hidden layer. This network attempts to fit the model

$\displaystyle f_k(x)= F\left\{w_{0k}^{(2)} + \sum_{j=1}^{J} w_{jk}^{(2)}F \left( w_{0j}^{(1)}
+ \sum_{i=1}^{I} w_{ij}^{(1)}x_i \right)\right\}
$

for the output unit $ y_k$. Feed-forward means that information can only flow forward from the input units to the first hidden layer, from the first hidden layer to the second hidden layer, and so on. Information cannot flow between the units of one layer.

Figure 8.2: Feed-forward network with one hidden layer.
\includegraphics[width=1.5\defpicwidth]{net.ps}