TF core API in R

We learned about the TensorFlow core API in Chapter 1. In R, this API is implemented with the tensorflow 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/tensorflow/articles/examples/mnist_softmax.html.

Note

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

1. First, load the library:

library(tensorflow)

2. Define the hyper-parameters:

batch_size <- 128
num_classes <- 10
steps <- 1000

3. Prepare the data:

datasets <- tf$contrib$learn$datasets
mnist <- datasets$mnist$read_data_sets("MNIST-data", one_hot = TRUE)

The data is loaded from the TensorFlow dataset library and is already normalized to fall into the [0,1] range.

4. Define the model:

# Create the model
x <- tf$placeholder(tf$float32, shape(NULL, 784L))
W <- tf$Variable(tf$zeros(shape(784L, num_classes)))
b <- tf$Variable(tf$zeros(shape(num_classes)))
y &lt...