Deep neural network (Theory)

What I have learnt will be focusing on Deep L-layer neural network. Deep neural network represent a network with lots of hidden layers instead of logistic regression or 2 layer NN. Some problem can be tackled easily by using a shallow neural network or logistic regression. It was suggested that we can start for a problem by using logistic regression or shallow neural network. Then, we can try to tune the number of hidden layer or some hyper-parameter to figure the most suitable deep neural network.