Neural networks and deep learning Chapter 4: feedforward neural networks reading questions

1. What are the main characteristics of Neural Networks ？

2. Why have activation functions ？ What are the characteristics of the activation function ？

3.logistic Functions and Tanh Similarities and differences of functions ？

4.ReLU Why can functions be widely used , What are the advantages and disadvantages of it ？

5. According to the general approximation theorem , Feedforward neural network can fit any function , What causes it to fail to achieve the desired effect ？

6. What is feature extraction

7. Let's deduce MLP Back propagation of

8. Why do gradients disappear and explode ？ How to reduce ？