Applied Deep Learning with Python

Dealing with New Data

The core idea of machine learning models—neural networks included—is that they can learn patterns from data. Imagine that a model was trained with a certain dataset and it is now issuing predictions. Now, imagine that new data is available. What strategies can we employ so that a model can take advantage of the newly available data to learn new patterns and improve its predictions?

In this section, we discuss two strategies: re-training an old model and training a new model.