A data scientist aims to improve a neural network model $($for classification$)$ that has

training accuracy $0\mathrm{.}99$ and testing accuracy $0\mathrm{.}6.$ Which strategy is MOST LIKELY

to improve the model?

A$.$ Add another hidden, dense layer.

B$.$ Apply dropout to all hidden layers, that is$,$ some activation will be randomly

set to $0$ during training.

C$.$ Use a smaller learning rate.

D$.$ Use a smaller training dataset