model.add_node(shift_layer, name="shift", input="sequence_input")
model.add_node(rnn_layer, name="rnn", input="shift")
model.add_node(TimeDistributedDense(input_dim), name="tdd", input="rnn")
model.add_node(Activation("softmax"), name="softmax", input="tdd")
model.add_output(name="output", input="softmax")
model.compile(loss={"output": "categorical_crossentropy"}, optimizer="adam")
for iteration in range(1, 200):
print()
print("-" * 50)
print("Iteration", iteration)
model.fit(
{"initial_value": i_train, "sequence_input": X_train, "output": y_train},
batch_size=BATCH_SIZE,
nb_epoch=1,
validation_data={"initial_value": i_val, "sequence_input": X_val},
show_accuracy=True,
)
###
# Select 10 samples from the validation set at random so we can visualize errors
for i in range(10):
rowi, rowX, rowy = i_val[np.array([i])], X_val[np.array([i])], y_val[np.array([i])]
proba = model.predict_classes({"initial_value": rowi, "sequence_input": rowX}, verbose=0)["output"]
preds = proba.argmax(axis=-1)
start = convertor.decode(rowX[0])
correct = convertor.decode(rowy[0])
guess = convertor.decode(preds[0], calc_argmax=False)
print("Start : ", start)
model.add_node(Dense(1024, activation='relu'),
name='layer4',
inputs=['n11_f', 'n22_f', 'n23_f', 'n24_f'],
merge_mode='concat')
model.add_node(Dense(10, activation='softmax'), name='layer5', input='layer4')
model.add_output(name='output1', input='layer5')
print 'Training....'
model.compile(loss={'output1': 'categorical_crossentropy'},
optimizer='adam',
metrics=['accuracy'])
model.fit({
'n00': x_train,
'output1': y_train
},
nb_epoch=nb_epoch,
batch_size=batch_size,
validation_split=0.3,
shuffle=True,
verbose=1)
#Model result:
loss_and_metrics = model.evaluate(x_train,
y_train,
batch_size=batch_size,
verbose=1)
print 'Done!'
print 'Loss: ', loss_and_metrics[0]
print ' Acc: ', loss_and_metrics[1]
#Saving Model
