Network Error

  • show all 0 hide
You are viewing the full content

Definition

In supervised learning, neural networks learn to approximate a given data set with the help of a teaching signal. During the training process, the network processes input exemplars with its current set of connection weights, resulting in a corresponding output signal. This output is compared to the teaching signal, representing the desired output. The network error is computed as the sum of squared differences between the actual and desired output for each output unit.

Connectionism

Neural Networks for Control