Advantages

• Learning models

• several ways of training a MLP
• backpropagation is the most common

• Universal function approximators

• Able to generalise to new data

• can accurately identify / model previously unseen examples