def euclideanSqDistance(inputs):
if (len(inputs) != 2):
raise 'oops'
output = K.mean(K.square(inputs[1] - inputs[0]), axis=-1)
output = K.expand_dims(output, 1)
return output
def call(self, x, mask=None):
uit = dot_product(x, self.W)
if self.bias:
uit += self.b
uit = K.tanh(uit)
ait = dot_product(uit, self.u)
# ait = K.dot(uit, self.u)
a = K.exp(ait)
# 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 ε 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())
a = K.expand_dims(a)
weighted_input = x * a
return K.sum(weighted_input, axis=1)
def call(self, x, mask=None):
# size of x :[batch_size, sel_len, attention_dim]
# size of u :[batch_size, attention_dim]
# uit = tanh(xW+b)
uit = K.tile(K.expand_dims(self.W, axis=0), (K.shape(x)[0], 1, 1))
uit = tf.matmul(x, uit)
uit = K.tanh(K.bias_add(uit, self.b))
ait = K.dot(uit, self.u)
ait = K.squeeze(ait, -1)
ait = K.exp(ait)
if mask is not None:
# Cast the mask to floatX to avoid float64 upcasting in theano
ait *= K.cast(mask, K.floatx())
ait /= K.cast(K.sum(ait, axis=1, keepdims=True) + K.epsilon(), K.floatx())
ait = K.expand_dims(ait)
weighted_input = x * ait
output = K.sum(weighted_input, axis=1)
return output
def dot_product(x, kernel):
"""
Wrapper for dot product operation, in order to be compatible with both
Theano and Tensorflow
Args:
x (): input
kernel (): weights
Returns:
"""
if K.backend() == 'tensorflow':
return K.squeeze(K.dot(x, K.expand_dims(kernel)), axis=-1)
else:
return K.dot(x, kernel)
def call(self, inputs, **kwargs):
if type(inputs) is list: # true label is provided with shape = [None, n_classes], i.e. one-hot code.
assert len(inputs) == 2
inputs, mask = inputs
else: # if no true label, mask by the max length of capsules. Mainly used for prediction
# compute lengths of capsules
x = K.sqrt(K.sum(K.square(inputs), -1))
# generate the mask which is a one-hot code.
# mask.shape=[None, n_classes]=[None, num_capsule]
mask = K.one_hot(indices=K.argmax(x, 1), num_classes=x.get_shape().as_list()[1])
# inputs.shape=[None, num_capsule, dim_capsule]
# mask.shape=[None, num_capsule]
# masked.shape=[None, num_capsule * dim_capsule]
masked = K.batch_flatten(inputs * K.expand_dims(mask, -1))
return masked
def atae_lstm_new(self):
input_content = Input(shape=(self.max_len, ))
input_aspect = Input(shape=(self.aspect_max_len, ))
###先将每个字进行embed，然后将aspect进行embed，然后根据content中字的个数重复，然后重复后的每个aspect embedding跟每个字的进行串联
content_embed = Embedding(input_dim=self.max_content_vocab_size,
output_dim=self.content_embed_dim)
aspect_embed = Embedding(input_dim=self.max_content_vocab_size,
output_dim=self.aspect_embed_dim)
content_embedding = content_embed(input_content)
content_embedding = SpatialDropout1D(0.2)(content_embedding)
aspect_embedding = aspect_embed(input_aspect)
##对aspect的字符串向量进行一个pooling 60*128=>1*128
aspect_embedding = AveragePooling1D(
pool_size=self.aspect_max_len)(aspect_embedding)
aspect_flatten = Flatten()(aspect_embedding)
repeat_aspect_embedding = RepeatVector(self.max_len)(aspect_flatten)
##将重复后的aspect和content字进行串联
input_concat = concatenate(
[content_embedding, repeat_aspect_embedding], axis=-1)
##再加个LSTM
if (self.is_cudnn):
hidden_vecs, state_h, _ = CuDNNLSTM(
self.lstm_units, return_sequences=True,
return_state=True)(input_concat)
else:
hidden_vecs, state_h, _ = LSTM(self.lstm_units,
return_sequences=True,
return_state=True)(input_concat)
concat = concatenate([hidden_vecs, repeat_aspect_embedding], axis=-1)
# apply attention mechanism
attend_weight = Attention()(concat)
attend_weight_expand = Lambda(lambda x: K.expand_dims(x))(
attend_weight)
attend_hidden = multiply([hidden_vecs, attend_weight_expand])
attend_hidden = Lambda(lambda x: K.sum(x, axis=1))(attend_hidden)
attend_hidden_dense = Dense(self.lstm_units)(attend_hidden)
last_hidden_dense = Dense(self.lstm_units)(state_h)
final_output = Activation('tanh')(add(
[attend_hidden_dense, last_hidden_dense]))
dense_layer = Dense(self.dense_units, activation='relu')(final_output)
output_layer = Dense(self.n_classes, activation='softmax')(dense_layer)
return Model([input_content, input_aspect], output_layer)
def call(self, inputs, training=None):
# inputs.shape=[None, input_num_capsule, input_dim_capsule]
# inputs_expand.shape=[None, 1, input_num_capsule, input_dim_capsule]
inputs_expand = K.expand_dims(inputs, 1)
# Replicate num_capsule dimension to prepare being multiplied by W
# inputs_tiled.shape=[None, num_capsule, input_num_capsule, input_dim_capsule]
inputs_tiled = K.tile(inputs_expand, [1, self.num_capsule, 1, 1])
# Compute `inputs * W` by scanning inputs_tiled on dimension 0.
# x.shape=[num_capsule, input_num_capsule, input_dim_capsule]
# W.shape=[num_capsule, input_num_capsule, dim_capsule, input_dim_capsule]
# Regard the first two dimensions as `batch` dimension,
# then matmul: [input_dim_capsule] x [dim_capsule, input_dim_capsule]^T -> [dim_capsule].
# inputs_hat.shape = [None, num_capsule, input_num_capsule, dim_capsule]
inputs_hat = K.map_fn(lambda x: K.batch_dot(x, self.W, [2, 3]), elems=inputs_tiled)
# Begin: Routing algorithm ---------------------------------------------------------------------#
# The prior for coupling coefficient, initialized as zeros.
# b.shape = [None, self.num_capsule, self.input_num_capsule].
b = tf.zeros(shape=[K.shape(inputs_hat)[0], self.num_capsule, self.input_num_capsule])
assert self.routings > 0, 'The routings should be > 0.'
for i in range(self.routings):
# c.shape=[batch_size, num_capsule, input_num_capsule]
c = tf.nn.softmax(b, dim=1)
# c.shape = [batch_size, num_capsule, input_num_capsule]
# inputs_hat.shape=[None, num_capsule, input_num_capsule, dim_capsule]
# The first two dimensions as `batch` dimension,
# then matmal: [input_num_capsule] x [input_num_capsule, dim_capsule] -> [dim_capsule].
# outputs.shape=[None, num_capsule, dim_capsule]
outputs = squash(K.batch_dot(c, inputs_hat, [2, 2])) # [None, 10, 16]
if i < self.routings - 1:
# outputs.shape = [None, num_capsule, dim_capsule]
# inputs_hat.shape=[None, num_capsule, input_num_capsule, dim_capsule]
# The first two dimensions as `batch` dimension,
# then matmal: [dim_capsule] x [input_num_capsule, dim_capsule]^T -> [input_num_capsule].
# b.shape=[batch_size, num_capsule, input_num_capsule]
b += K.batch_dot(outputs, inputs_hat, [2, 3])
# End: Routing algorithm -----------------------------------------------------------------------#
return outputs
def call(self, x, mask=None):
features_dim = self.features_dim
step_dim = self.step_dim
eij = K.reshape(K.dot(K.reshape(x, (-1, features_dim)),
K.reshape(self.W, (features_dim, 1))), (-1, step_dim))
if self.bias:
eij += self.b
eij = K.tanh(eij)
a = K.exp(eij)
if mask is not None:
a *= K.cast(mask, K.floatx())
a /= K.cast(K.sum(a, axis=1, keepdims=True) + K.epsilon(), K.floatx())
a = K.expand_dims(a)
weighted_input = x * a
return K.sum(weighted_input, axis=1)
def expand_label_input(x):
x = K.expand_dims(x, axis=1)
x = K.expand_dims(x, axis=1)
x = K.tile(x, [1, 32, 32, 1])
return x
def call(self, inputs):
if K.dtype(inputs) != 'int32':
inputs = K.cast(inputs, 'int32')
out = K.gather(self.embeddings, inputs)
mask = K.expand_dims(K.clip(K.cast(inputs, 'float32'), 0, 1), axis=-1)
return out * mask
def splitter(y_true):
payoffs = y_true[:, 1]
payoffs = K.expand_dims(payoffs, 1)
y_true = y_true[:, 0]
y_true = K.expand_dims(y_true, 1)
return y_true, payoffs
def dot_product(x, kernel):
if K.backend() == 'tensorflow':
return K.squeeze(K.dot(x, K.expand_dims(kernel)), axis=-1)
else:
return K.dot(x, kernel)
def cabasc(self):
def sequence_mask(sequence):
return K.sign(K.max(K.abs(sequence), 2))
def sequence_length(sequence):
return K.cast(K.sum(sequence_mask(sequence), 1), tf.int32)
input_text = Input(shape=(self.max_len, ))
input_text_l = Input(shape=(self.max_len, ))
input_text_r = Input(shape=(self.max_len, ))
input_aspect = Input(shape=(1, ))
input_mask = Input(shape=(self.max_len, ))
word_embedding = Embedding(input_dim=self.max_content_vocab_size,
output_dim=self.content_embed_dim)
text_embed = SpatialDropout1D(0.2)(word_embedding(input_text))
text_l_embed = SpatialDropout1D(0.2)(word_embedding(input_text_l))
text_r_embed = SpatialDropout1D(0.2)(word_embedding(input_text_r))
asp_embedding = Embedding(input_dim=self.max_aspect_vocab_size,
output_dim=self.aspect_embed_dim)
aspect_embed = asp_embedding(input_aspect)
aspect_embed = Flatten()(aspect_embed) # reshape to 2d
# regarding aspect string as the first unit
hidden_l = GRU(self.lstm_units,
go_backwards=True,
return_sequences=True)(text_l_embed)
hidden_r = GRU(self.lstm_units, return_sequences=True)(text_r_embed)
# left context attention
context_attend_l = TimeDistributed(Dense(
1, activation='sigmoid'))(hidden_l)
# Note: I couldn't find `reverse_sequence` in keras
context_attend_l = Lambda(lambda x: tf.reverse_sequence(
x, sequence_length(x), 1, 0))(context_attend_l)
context_attend_l = Lambda(lambda x: K.squeeze(x, -1))(context_attend_l)
# right context attention
context_attend_r = TimeDistributed(Dense(
1, activation='sigmoid'))(hidden_r)
context_attend_r = Lambda(lambda x: K.squeeze(x, -1))(context_attend_r)
# combine context attention
# aspect_text_embed = subtract([add([text_l_embed, text_r_embed]), text_embed])
# aspect_text_mask = Lambda(lambda x: sequence_mask(x))(aspect_text_embed)
# text_mask = Lambda(lambda x: sequence_mask(x))(text_embed)
# context_mask = subtract([text_mask, aspect_text_mask])
# aspect_text_mask_half = Lambda(lambda x: x*0.5)(aspect_text_mask)
# combine_mask = add([context_mask, aspect_text_mask_half]) # 1 for context, 0.5 for aspect
context_attend = multiply(
[add([context_attend_l, context_attend_r]), input_mask])
# apply context attention
context_attend_expand = Lambda(lambda x: K.expand_dims(x))(
context_attend)
memory = multiply([text_embed, context_attend_expand])
# sentence-level content attention
sentence = Lambda(lambda x: K.mean(x, axis=1))(memory)
final_output = ContentAttention()([memory, aspect_embed, sentence])
dense_layer = Dense(self.dense_units, activation='relu')(final_output)
output_layer = Dense(self.n_classes, activation='softmax')(dense_layer)
return Model(
[input_text, input_text_l, input_text_r, input_aspect, input_mask],
output_layer)
