A data scientist aims to improve a neural network model $($for classification$)$ that has training accuracy $0\mathrm{.}58$ and testing accuracy $0\mathrm{.}55.$ Which strategy is MOST LIKELY to improve the model?

$$a$)$ Add two more hidden, dense layers.

$$b$)$ Add a regularization term to loss function.

$$c$)$ Use a smaller training dataset.

d$)$ Use a larger learning rate.