Exercise

Visualizing the prototypes

Now, we are going to compute the centroids of the coordinates for each of the two archetypes for each label. We did something similar before for the t-SNE embedding. The goal is to have a representation or prototype of each label in this new two-dimensional space.

The ggplot2 and data.table packages are pre-loaded, as well as the X_matrix object and the fashion_mnist dataset.

Instructions 1/3

undefined XP

1

2

3

Create three new columns in the data table X_matrix with the following information: the label from fashion_mnist in numeric format, the mean of Arch1 and the mean of Arch2 for each label.