Hands-On Python Deep Learning for the Web

The importance of using pre-trained models

The use of pre-trained models is commonly referred to as transfer learning. Transfer learning is not something that is very fundamental to deep learning and it is just a methodology. It doesn't denote a particular deep learning model but its implications of transfer learning are very effective, especially in a deep learning context.

We human beings do not learn each and every task from scratch; we try to utilize our past experiences to do tasks that are similar in nature. This is transfer learning. We tend to transfer the knowledge of our past experiences to similar tasks that we are met with.

But how is this applicable to deep learning? Let's find out.

When a neural network is trained for a particular ...