Deep Learning with Microsoft Cognitive Toolkit Quick Start Guide

Picking a loss function

When we have the structure for the model, it is time to take a look at how to optimize it. For this, we need a loss function to minimize. There are quite a few loss functions to choose from.

The right loss function depends on what kind of problem you are solving. For example, in a classification model like ours, we need a loss function that can measure the difference between a predicted class and an actual class. It needs to do so for three classes. The categorical cross entropy function is a good candidate. In CNTK, this loss function is implemented as cross_entropy_with_softmax:

label = input_variable(3)loss = cross_entropy_with_softmax(z, label)

We need to import the cross_entropy_with_softmax function from the ...