attention層的定義：（思路參考https://github.com/philipperemy/keras-attention-mechanism）
# Attention GRU network
class AttLayer(Layer):
def __init__(self, **kwargs):
self.init = initializations.get('normal')
#self.input_spec = [InputSpec(ndim=3)]
super(AttLayer, self).__init__(**kwargs)
def build(self, input_shape):
assert len(input_shape)==3
#self.W = self.init((input_shape[-1],1))
self.W = self.init((input_shape[-1],))
#self.input_spec = [InputSpec(shape=input_shape)]
self.trainable_weights = [self.W]
super(AttLayer, self).build(input_shape) # be sure you call this somewhere!
def call(self, x, mask=None):
eij = K.tanh(K.dot(x, self.W))
ai = K.exp(eij)
weights = ai/K.sum(ai, axis=1).dimshuffle(0,'x')
weighted_input = x*weights.dimshuffle(0,1,'x')
return weighted_input.sum(axis=1)
def get_output_shape_for(self, input_shape):
return (input_shape[0], input_shape[-1])
具體的用法：
input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')
embedded_sequences = embedding_layer(input)
l_lstm = Bidirectional(LSTM(100, return_sequences=True))(embedded_sequences)
l_att = AttLayer()(l_lstm)
preds = Dense(2, activation='softmax')(l_att)
model = Model(sequence_input, preds)
model.compile(loss='categorical_crossentropy',
optimizer='rmsprop',
metrics=['acc'])
print("model fitting - attention GRU network")
model.summary()
model.fit(x_train, y_train, validation_data=(x_val, y_val),
nb_epoch=10, batch_size=50)
