from keras import backend as K
from keras.engine.topology import Layer
from keras import initializers, regularizers, constraints

class Attention_layer(Layer):
    """
        Attention operation, with a context/query vector, for temporal data.
        Supports Masking.
        Follows the work of Yang et al. [https://www.cs.cmu.edu/~diyiy/docs/naacl16.pdf]
        "Hierarchical Attention Networks for Document Classification"
        by using a context vector to assist the attention
        # Input shape
            3D tensor with shape: `(samples, steps, features)`.
        # Output shape
            2D tensor with shape: `(samples, features)`.
        :param kwargs:
        Just put it on top of an RNN Layer (GRU/LSTM/SimpleRNN) with return_sequences=True.
        The dimensions are inferred based on the output shape of the RNN.
        Example:
            model.add(LSTM(64, return_sequences=True))
            model.add(AttentionWithContext())
        """

    def __init__(self,
                 W_regularizer=None, b_regularizer=None,
                 W_constraint=None, b_constraint=None,
                 bias=True, **kwargs):

        self.supports_masking = True
        #self.init = initializations.get('glorot_uniform')
        self.init = initializers.get('glorot_uniform')
        self.W_regularizer = regularizers.get(W_regularizer)
        self.b_regularizer = regularizers.get(b_regularizer)

        self.W_constraint = constraints.get(W_constraint)
        self.b_constraint = constraints.get(b_constraint)

        self.bias = bias
        super(Attention_layer, self).__init__(**kwargs)

    def build(self, input_shape):
        assert len(input_shape) == 3

        self.W = self.add_weight((input_shape[-1], input_shape[-1],),
                                 initializer=self.init,
                                 name='{}_W'.format(self.name),
                                 regularizer=self.W_regularizer,
                                 constraint=self.W_constraint)
        if self.bias:
            self.b = self.add_weight((input_shape[-1],),
                                     initializer='zero',
                                     name='{}_b'.format(self.name),
                                     regularizer=self.b_regularizer,
                                     constraint=self.b_constraint)

        super(Attention_layer, self).build(input_shape)

    def compute_mask(self, input, input_mask=None):
        # do not pass the mask to the next layers
        return None

    def call(self, x, mask=None):
        uit = K.dot(x, self.W)

        if self.bias:
            uit += self.b

        uit = K.tanh(uit)

        a = K.exp(uit)

        # apply mask after the exp. will be re-normalized next
        if mask is not None:
            # Cast the mask to floatX to avoid float64 upcasting in theano
            a *= K.cast(mask, K.floatx())

        # in some cases especially in the early stages of training the sum may be almost zero
        # and this results in NaN's. A workaround is to add a very small positive number to the sum.
        # a /= K.cast(K.sum(a, axis=1, keepdims=True), K.floatx())
        a /= K.cast(K.sum(a, axis=1, keepdims=True) + K.epsilon(), K.floatx())
        weighted_input = x * a
        return K.sum(weighted_input, axis=1)

    def get_output_shape_for(self, input_shape):
        return input_shape[0], input_shape[-1]

 

step-1 keras定義自己的層的方法

要定義自己的層，要實現下面的三個方法(來自官方文件)

1. build(input_shape): 這是你定義權重的地方。這個方法必須設self.built = True，可以通過呼叫super([Layer], self).build()完成。
2. call(x): 這裡是編寫層的功能邏輯的地方。你只需要關注傳入call的第一個引數：輸入張量，除非你希望你的層支援masking。call(x): 這裡是編寫層的功能邏輯的地方。你只需要關注傳入call的第一個引數：輸入張量，除非你希望你的層支援masking。
3. compute_output_shape(input_shape)
   : 如果你的層更改了輸入張量的形狀，你應該在這裡定義形狀變化的邏輯，這讓Keras能夠自動推斷各層的形狀。compute_output_shape(input_shape): 如果你的層更改了輸入張量的形狀，你應該在這裡定義形狀變化的邏輯，這讓Keras能夠自動推斷各層的形狀。