TF estimator API in R

We learned about the TensorFlow estimator API in Chapter 2. In R, this API is implemented with the tfestimator R package. 

As an example, we provide a walkthrough of the MLP Model for classifying handwritten digits from the MNIST dataset at the following link: https://tensorflow.rstudio.com/tfestimators/articles/examples/mnist.html.

Note

You can follow along with the code in the Jupyter R notebook ch-17b_TFEstimator_in_R.

1. First, load the libraries:

library(tensorflow)
library(tfestimators)

2. Define the hyper-parameters:

batch_size <- 128
n_classes <- 10
n_steps <- 100

3. Prepare the data:

# initialize data directory
data_dir <- "~/datasets/mnist"
dir.create(data_dir, recursive = TRUE, showWarnings = FALSE)

# download the MNIST data sets, and read them into R
sources <- list(
  train = list(
    x = "https://storage.googleapis.com/cvdf-datasets/mnist/train-images-idx3-ubyte.gz",
    y = "https://storage.googleapis.com/cvdf-datasets/mnist/train-labels-idx1-ubyte.gz...