Tensorflow – The R session continues to abort when using Keras to form a neural network

I'm trying to create a neural network capable of performing a binary classification on a dataset. However, when I try to use the fit function with my template, a pop-up window saying "R Aborted Session canceled .R encountered a fatal error." The session was terminated. "

What could it crash?

The dataset consists of 5 columns on 28,200 rows. With a column being the target variable.

I've set tensorflow on version 1.4, because the most recent versions apparently cause the crash of older computers such as mine, but that did not solve the problem.

In addition, I made it crash when I specified the layers, but I changed them and it did not crash until I tried to adapt the model.





x_train <- as.matrix (training_set[,2:5])
y_train <- as.matrix (training_set[,1])
x_test <- as.matrix (test_set[,2:5])
y_test <- as.matrix (test_set[,1])


model <- keras_model_sequential()
model %>%
layer_dense (units = 32, activation = "relu", input_shape = c (4))%>%
layer_dense (units = 1, activation = & # 39; sigmoid & # 39;)


model%>% compile (
loss = "binary_crossentropy",
optimizer = "rmsprop",
metric = c (& # 39;)
)

%>% fit template (
x_train, y_train,
batch_size = 32,
times = 10,
validation_data = list (x_test, y_test)
)