May 2018
Beginner to intermediate
384 pages
10h 19m
English
In the Gradient-based learning section of this chapter, we established the equations for weight and bias updates for the deep neural network as follows:


In these equations, the learning rate is denoted by
. The learning rate for the gradient descent algorithm defines the size of the step that algorithm takes with each training set instance. If the learning rate is too high, the average loss across the gradient descent steps will ...
Read now
Unlock full access