[R] Using Keras on R

텐서플로우가 뒷단에서 돌아가는 Keras는 사용하기가 미친듯이 편리하게 되어있다.

출처 : https://tensorflow.rstudio.com/keras/

library(keras)

install_keras()

mnist <- dataset_mnist()

x_train <- mnist$train$x

y_train <- mnist$train$y

x_test <- mnist$test$x

y_test <- mnist$test$y

summary(x_train)

head(x_train)

dim(x_train)

x_train <- array_reshape(x_train, c(nrow(x_train), 784))

x_test <- array_reshape(x_test, c(nrow(x_test), 784))

x_train  <- x_train /  255

x_test <- x_test / 255

y_train <- to_categorical(y_train, 10)

y_test <- to_categorical(y_test, 10)

model <- keras_model_sequential() 

model %>% 

  layer_dense(units = 256, activation = 'relu', input_shape = c(784)) %>% 

  layer_dropout(rate = 0.4) %>% 

  layer_dense(units = 128, activation = 'relu') %>%

  layer_dropout(rate = 0.3) %>%

  layer_dense(units = 10, activation = 'softmax')

summary(model)

model %>% compile(

  loss = 'categorical_crossentropy',

  optimizer = optimizer_rmsprop(),

  metrics = c('accuracy')

)

history <- model %>% fit(

  x_train, y_train, 

  epochs = 30, batch_size = 128, 

  validation_split = 0.2

)