###Fitting RNNs using Keras:
### Taken/adapted from "Deep Learning wth R" by Chollet


library(keras)

## Using IMDB dataset in Keras:
max_features <- 10000                                             
maxlen <- 500                                                     

cat("Loading data...\n")
imdb <- dataset_imdb(num_words = max_features)
c(c(input_train, y_train), c(input_test, y_test)) %<-% imdb
cat(length(input_train), "train sequences\n")
#25000 train sequences
cat(length(input_test), "test sequences")
#25000 test sequences
cat("Pad sequences (samples x time)\n")
input_train <- pad_sequences(input_train, maxlen = maxlen)
input_test <- pad_sequences(input_test, maxlen = maxlen)
cat("input_train shape:", dim(input_train), "\n")
#input_train shape: 25000 500 
cat("input_test shape:", dim(input_test), "\n")
#input_test shape: 25000 500

## RNN:
model <- keras_model_sequential() %>%
  layer_embedding(input_dim = max_features, output_dim = 32) %>%
  layer_simple_rnn(units = 32) %>%
  layer_dense(units = 1, activation = "sigmoid")

model %>% compile(
  optimizer = "rmsprop",
  loss = "binary_crossentropy",
  metrics = c("acc")
)

history <- model %>% fit(
  input_train, y_train,
  epochs = 10,
  batch_size = 128,
  validation_split = 0.2
)

## LSTM:
model <- keras_model_sequential() %>%
  layer_embedding(input_dim = max_features, output_dim = 32) %>%
  layer_lstm(units = 32) %>%
  layer_dense(units = 1, activation = "sigmoid")

model %>% compile(
  optimizer = "rmsprop",
  loss = "binary_crossentropy",
  metrics = c("acc")
)

history <- model %>% fit(
  input_train, y_train,
  epochs = 10,
  batch_size = 128,
  validation_split = 0.2
)

##1D CNN:
model <- keras_model_sequential() %>%
  layer_embedding(input_dim = max_features, output_dim = 128,
                  input_length = maxlen) %>% 
  layer_conv_1d(filters = 32, kernel_size = 7, activation = "relu") %>%
  layer_max_pooling_1d(pool_size = 5) %>%
  layer_conv_1d(filters = 32, kernel_size = 7, activation = "relu") %>%
  layer_global_max_pooling_1d() %>%
  layer_dense(units = 1)
summary(model)
model %>% compile(
  optimizer = optimizer_rmsprop(lr = 1e-4),
  loss = "binary_crossentropy",
  metrics = c("acc")
)
history <- model %>% fit(
  input_train, y_train,
  epochs = 10,
  batch_size = 128,
  validation_split = 0.2
)

##1D CNN + RNN:
model <- keras_model_sequential() %>%
  layer_embedding(input_dim = max_features, output_dim = 128,
                  input_length = maxlen) %>%
  layer_conv_1d(filters = 32, kernel_size = 7, activation = "relu") %>%
  layer_max_pooling_1d(pool_size = 5) %>%
  layer_conv_1d(filters = 32, kernel_size = 7, activation = "relu") %>%
  layer_gru(units=32) %>%
  layer_dense(units = 1)
summary(model)
model %>% compile(
  optimizer = optimizer_rmsprop(lr = 1e-4),
  loss = "binary_crossentropy",
  metrics = c("acc")
)
history <- model %>% fit(
  input_train, y_train,
  epochs = 10,
  batch_size = 128,
  validation_split = 0.2
)

###FNN + RNN:
model <- keras_model_sequential() %>%
  layer_embedding(input_dim = max_features, output_dim = 32,
     input_length = maxlen) %>%
  layer_dense(units =16, activation="relu"   ) %>%
  layer_dense(units =16, activation="relu" ) %>%

  layer_gru(units = 32) %>%
  layer_dense(units = 1, activation = "sigmoid")

model %>% compile(
  optimizer = "rmsprop",
  loss = "binary_crossentropy",
  metrics = c("acc")
)

history <- model %>% fit(
  input_train, y_train,
  epochs = 10,
  batch_size = 128,
  validation_split = 0.2
)