Training a feedforward neural network to predict glucose levels

In this chapter, we aim to predict the next n values of the glucose level based on the previous m values via a fully connected feedforward neural network. To do this, we will do the following:

1. Introduce KNIME Deep Learning – Keras Integration.
2. Design the fully connected feedforward neural architecture.
3. Train the model.
4. Create an alarm that alerts the end user when glucose levels are predicted to exceed the safety threshold.

Let’s start with KNIME Deep Learning – Keras Integration.

KNIME Deep Learning Keras Integration

In this section, we will install and set up KNIME Deep Learning – Keras Integration to train neural networks in KNIME Analytics Platform.

KNIME Deep Learning – Keras Integration wraps Keras functions into KNIME nodes, giving them a configuration dialog to set the parameters. Such nodes allow you to read, write, build, train, and execute...