LSTM Network with Keras

#------------- DEEP LEARNING -----------------#

#-----LSTM Network with Keras------#

#--- Parameters----#

# encodes input sequence dense vectors 
embed_dim = 128

# transforms the vector sequence into a single vector
lstm_out = 200

# batch size of 32 is a good starting point
batch_size = 32

# epochs
nb_epoch = 10

#------# Build the LSTM model #-----------------#

print('Building model...') 

# Initialising the RNN
model = Sequential()

#adding an input layer and the first hidden layer
model.add(Embedding(2500, embed_dim, 
                    input_length = features.shape[1], 
                    dropout = 0.2)) 
# Adding the second hidden layer
model.add(LSTM(lstm_out, dropout_U = 0.2, dropout_W = 0.2))
# Adding the output layer
model.add(Dense(2, activation='softmax'))

# Compile model
model.compile( optimizer='adam', # optimazer
              loss = 'categorical_crossentropy', # loss function
              metrics = ['accuracy']) # list of metrics

model.name = 'LSTM model'
print(model.summary())

# Output:

Building model...
Model: "LSTM model"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
embedding_1 (Embedding)      (None, 39, 128)           320000    
_________________________________________________________________
lstm_1 (LSTM)                (None, 200)               263200    
_________________________________________________________________
dense_1 (Dense)              (None, 2)                 402       
=================================================================
Total params: 583,602
Trainable params: 583,602
Non-trainable params: 0
_________________________________________________________________
None