R Deep Learning Cookbook
上QQ阅读APP看书,第一时间看更新

Getting ready

The ANN is a collection of artificial neurons that perform simple operations on the data; the output from this is passed to another neuron. The output generated at each neuron is called its activation function. An example of a multilayer perceptron model can be seen in the following screenshot:

An example of a multilayer neural network

Each link in the preceding figure is associated to weights processed by a neuron. Each neuron can be looked at as a processing unit that takes input processing and the output is passed to the next layer, as shown in the following screenshot:

An example of a neuron getting three inputs and one output

The preceding figure demonstrates three inputs combined at neuron to give an output that may be further passed to another neuron. The processing conducted at the neuron could be a very simple operation such as the input multiplied by weights followed by summation or a transformation operation such as the sigmoid activation function.