Reference Work Entry

Encyclopedia of Neuroscience

pp 2548-2548

Network Error


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.


Neural Networks for Control

You are viewing the full content