IN THE CLASS ROOM WE WERE TAUGHT TO USE NEURAL NETWORKS FOR CLASSIFICATION PURPOSE, BUT IN OPTIONAL LAB FOR MODEL EVALUATION AND SELECTION ITS BEING USED FOR REGRESSION. How come?
Neural Networks can be used for any problem. The difference is in only the last layer where for the classification problems, we use the sigmoid or softmax activation function whereas, for regression problems, ReLU is more common.
Best,
Saif.