Problem
A. How do neural networks differ from logistic regression?
B. If we have three classes of outputs in the final layer of a neural network, how many weight vectors do we need to train in the final layer?
C. Say that our input to an activation is -3. Show the output for the sigmoid, hyperbolic tangent, ReLU, and softplus activation functions.
D. What is the difference in the output layer between a neural network used for classification, and one used for regression?
E. Describe why we need to use regularization in neural networks.