Hands-On Machine Learning on Google Cloud Platform

Building the model

Our aim is to use data in the dataset to make predictions. In particular, we want to predict the presence of carbon dioxide in the air based on the data available in the .csv file. We need input and output to train and test our network. It is clear that the input is represented by the data present in the dataset. We must then construct our output; we will do so by supposing we want to predict the CO2 present in the atmosphere at time t + 1 with respect to the value measured at time t. So we will have:

Input = data(t)

Output = data(t + 1)

We have said that a recurrent network has memory, and it is maintained by fixing the so-called time step. The time step has to do with how many steps back in time backprop uses when calculating ...