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$)$ Use a smaller training dataset.

$$c$)$ Apply dropout to all hidden layers, that is$,$ some activation will be randomly set to $0$ during training.

$$d$)$ Use a smaller learning rate.