In AlfonsoRReyes/rDeepThought:

Source: https://cran.r-project.org/web/packages/kerasR/vignettes/introduction.html

# images are black and white
library(kerasR)

mnist <- load_mnist()
#  increase the number of dimension in the X_train tensor by one
# additional dimension could represent a “gray” channel.
X_train <- array(mnist$X_train, dim = c(dim(mnist$X_train), 1)) / 255
Y_train <- to_categorical(mnist$Y_train, 10)
X_test <- array(mnist$X_test, dim = c(dim(mnist$X_test), 1)) / 255
Y_test <- mnist$Y_test

#  build a CNN model by using the convolution specific Conv2D and MaxPooling layers
# deep convolution layers followed by Dense layers at the top of the network
mod <- Sequential()

mod$add(Conv2D(filters = 32, kernel_size = c(3, 3),
               input_shape = c(28, 28, 1)))
mod$add(Activation("relu"))
mod$add(Conv2D(filters = 32, kernel_size = c(3, 3),
               input_shape = c(28, 28, 1)))
mod$add(Activation("relu"))
mod$add(MaxPooling2D(pool_size=c(2, 2)))
mod$add(Dropout(0.25))

# The flatten layer converts its inputs into a 2-dimensional tensor 
mod$add(Flatten())
mod$add(Dense(128))
mod$add(Activation("relu"))
mod$add(Dropout(0.25))
mod$add(Dense(10))
mod$add(Activation("softmax"))

# compile it and fit it to the data
keras_compile(mod,  loss = 'categorical_crossentropy', optimizer = RMSprop())
keras_fit(mod, X_train, Y_train, batch_size = 32, epochs = 5, verbose = 1,
          validation_split = 0.1)

Y_test_hat <- keras_predict_classes(mod, X_test)
table(Y_test, Y_test_hat)
mean(Y_test == Y_test_hat)

AlfonsoRReyes/rDeepThought documentation built on May 3, 2019, 6:42 p.m.