Neural networks in machine learning
This is not a book about neural networks. But since the tools we’ve introduced (KIE and Python notebooks) also allow you to work more effectively with neural networks, it’s worth having a simple introduction to them.
To see the problem neural networks can help solve, imagine we took our sales graph (Figure 11.4 from Chapter 11) and added the sales data for our chocolate crunch bars. We might end up with a graph such as Figure 12.3:
Figure 12.3 – More complicated sales data
This graph is more difficult to describe – while we’ve highlighted our Chocolate Crunch sales with red circles, the group in the top right is mixed in with Silk Tray orders. There is no single sentence we can use to describe the pattern and there is no extra feature (or dimension) that we can use to clarify this situation. It is for processing complex patterns like this that neural networks work best.
Reality...
