3.1. Which of the following are some common issues we experience while training deep

neural networks?

As the data gets big and complex, convergence might take lot of time as the training gets

slow.

O

As the network gets deeper i.e., by adding more layers, the magnitude of the gradients

might below too less or high, which might lead to vanishing or exploding gradients.

O

Since deep learning models are complex by nature, generally we observe under-fitting

issue.