Exercise

Reducing data with GLRM

We are going to reduce the dimensionality of the fashion MNIST sample data using the GLRM implementation of h2o.

In order to do this, in the next steps we are going to:

Start a connection to a h2o cluster by invoking the method h2o.init().
Store the fashion_mnist data into the h2o cluster with as.h2o().
Launch a GLRM model with K=2 (rank-2 model) using the h2o.glrm() function.

As we have discussed in the video session, it is important to check the convergence of the objective function. Note that here we are also fixing the seed to ensure the same results.

The h2o package and fashion_mnist data are pre-loaded in the environment.

Instructions 1/2

XP

1
2

Start the h2o cluster.
Store the fashion_mnist dataset into the h2o format.