Forwardpropagating step

Figure 2 shows a general appearance of a neuron with its connections. Each connection from ith to the jth neuron is associated with a quantity called weight or connection strength (wj). A net input (called activation) for each neuron is the sum of all its input values multiplied by their corresponding connection weights, expressed as ai = xiwji+° i where i the total of neurons in the previous layer and Oj is a bias term which influences the horizontal offset of the function (fixed value of 1). Once the activation of a neuron is calculated, we can determine the output value (i.e., the response) by applying a transfer function:

Many transfer functions may be used, for example, a linear function, a threshold function, a sigmoid function, etc. (Figure 3). A sigmoid function is often used, because it has nonlinearity, which is given by

The weights play an important role in the propagation of the signal in the network. They establish a link between input pattern and its associated output pattern, that is, they contain the knowledge of the neural network about the problem-solution relation.

The forward-propagation step begins with the presentation of an input pattern to the input layer, and continues as activation-level calculations propagate forward till the output layer through the hidden layer(s). In each

Figure 2 Basic processing element (neuron) in a network. Each input connection value (x,) is associated with a weight (wj). The output value (X,- = f(aj)) can fan out to another unit.
Solar Power

Solar Power

Start Saving On Your Electricity Bills Using The Power of the Sun And Other Natural Resources!

Get My Free Ebook

Post a comment