Question

Q. What is not a good weight initialization idea for deep neural Networks?

A. Initialize all the weights with 0
B. Initialize all weights with small random
numbers scaled by 0.1
c. Initialize all the weights with random numbers scaled by 100
D. None of the options .
​

Answer 1

Explanation:

The aim of weight initialization is to prevent layer activation outputs from exploding or vanishing during the course of a forward pass through a deep neural network.