Exercise

Your first neural net

All that buzz around AI and Deep Learning, but we already promised you it is not so scary when you take a look under the hood.

You're going to create your first neural network, to solve a classification task.

You are working with labelled sensor data which has:

  • 16 input columns, representing 16 features coming from a movement sensor of a smartwatch.
  • a target column representing one of 4 movement types: walking, running, sitting, lying.

All the ingredients have been loaded for you: the Sequential() model and the Dense() layer -- you just have to put the ingredients in the right order, just like stacking pancakes!

Instructions

100 XP
  • Initialize a Sequential() network.
  • Set a fully connected Dense() hidden layer with 8 units (neurons), making sure to specify the correct input size to match the dimensions of the input data.
  • Set another fully connected Dense() layer at the output, making sure to specify the correct number of output units, defined by the number of output classes of your problem.
  • Compile the model!