Java Deep Learning Cookbook

How it works...

For step 1, we need to make sure that nOut() for the preceding layer should have the same number of neurons as nIn() for the output layer.

So, incomingConnectionCount should be the same as outgoingConnectionCount from the preceding layer.

We discussed the SOFTMAX activation function earlier in Chapter 1, Introduction to Deep Learning in Java. Our use case (customer churn) is an example for the binary classification model. We are looking for a probabilistic outcome, that is, the probability of a customer being labeled happy or unhappy, where 0 represents a happy customer and 1 represents an unhappy customer. This probability will be evaluated, and the neural network will train itself during the training process.

The proper ...