In this chapter we will explore how to train and build deep prediction models. We will focus on feedforward neural networks, which are perhaps the most common type and a good starting point.
This chapter will cover the following topics:
Getting started with deep feedforward neural networks
Common activation functions: rectifiers, hyperbolic tangent, and maxout
Picking hyperparameters
Training and predicting new data from a deep neural network
Use case – training a deep neural network for automatic classification
In this chapter, we will not use any new packages. The only requirements are to source the checkpoint.R
file to set up the R environment for the rest of the code shown and to initialize the H2O cluster. Both can be done using the following code:
source("checkpoint.R") options(width = 70, digits = 2) cl <- h2o.init( max_mem_size = "12G", nthreads = 4)