Create MICNN.py
This commit is contained in:
missQian
GitHub
parent
8d908bcffa
commit
f653d53e04
@ -0,0 +1,216 @@
|
||||
import tensorflow as tf
|
||||
from tensorflow.contrib.layers import fully_connected
|
||||
from sklearn.metrics import f1_score, precision_score, recall_score
|
||||
from preprocess import *
|
||||
|
||||
class MICNN():
|
||||
def __init__(self, params, vocabs, my_embeddings=None):
|
||||
self.params = params
|
||||
self.vocabs = vocabs
|
||||
for key in params:
|
||||
setattr(self, key, params[key])
|
||||
if self.pretrain:
|
||||
self.my_embeddings = my_embeddings
|
||||
|
||||
def build_embedding(self):
|
||||
if self.pretrain:
|
||||
embedding_placeholder = tf.placeholder(tf.float32,
|
||||
[len(self.vocabs), self.embedding_size])
|
||||
else:
|
||||
embedding_placeholder = tf.get_variable("embedding", initializer=tf.random_uniform(
|
||||
[len(self.vocabs), self.embedding_size], -1, 1), dtype=tf.float32)
|
||||
|
||||
return embedding_placeholder
|
||||
|
||||
|
||||
def build(self):
|
||||
tf.reset_default_graph()
|
||||
# length of each sentence in the whole batch
|
||||
self.sequence_length = tf.placeholder(tf.int64, [None])
|
||||
self.article_lens = tf.placeholder(tf.int64, [None])
|
||||
|
||||
# input data is in form of [batch_size, article_len, sentence_len]
|
||||
self.train_inputs = tf.placeholder(tf.int32, shape=[None, None, None])
|
||||
self.embedding_placeholder = self.build_embedding()
|
||||
self.embed = tf.nn.embedding_lookup(self.embedding_placeholder, self.train_inputs)
|
||||
self.output = tf.placeholder(tf.int64, [None])
|
||||
self.keep_prob = tf.placeholder(tf.float32)
|
||||
|
||||
shape = tf.shape(self.embed)
|
||||
|
||||
# the inputs are reshaped to [all sentences, sentence_len] to be passed to LSTM
|
||||
embed = tf.reshape(self.embed, [shape[0] * shape[1], shape[2], self.embedding_size])
|
||||
self.sequence_length = tf.reshape(self.sequence_length, [tf.shape(embed)[0]])
|
||||
|
||||
f_cell = tf.contrib.rnn.GRUCell(num_units=self.hidden_size)
|
||||
f_cell_drop = tf.contrib.rnn.DropoutWrapper(f_cell, input_keep_prob=self.keep_ratio)
|
||||
self.f_network = tf.contrib.rnn.MultiRNNCell([f_cell_drop] * self.num_layers)
|
||||
|
||||
b_cell = tf.contrib.rnn.GRUCell(num_units=self.hidden_size, reuse=False )
|
||||
b_cell_drop = tf.contrib.rnn.DropoutWrapper(b_cell, input_keep_prob=self.keep_ratio)
|
||||
self.b_network = tf.contrib.rnn.MultiRNNCell([b_cell_drop] * self.num_layers)
|
||||
|
||||
bi_outputs, bi_states = tf.nn.bidirectional_dynamic_rnn(self.f_network, self.b_network,
|
||||
embed, dtype=tf.float32,
|
||||
sequence_length=self.sequence_length)
|
||||
fw_states, bw_states = bi_states
|
||||
state = tf.concat([fw_states, bw_states], 2)
|
||||
state = tf.reshape(state, [shape[0], shape[1], 2 * self.hidden_size, 1])
|
||||
|
||||
# CNN on top of Bi-LSTM to get the article features
|
||||
art_pooled_outputs = list()
|
||||
|
||||
for i, art_filter_size in enumerate(self.art_filter_sizes):
|
||||
filter_shape = [art_filter_size, 2 * self.hidden_size, 1, self.art_num_filters]
|
||||
art_b = tf.Variable(tf.constant(0.1, shape=[self.art_num_filters]))
|
||||
art_W = tf.Variable(tf.truncated_normal(filter_shape, stddev=0.1), name="art_W")
|
||||
|
||||
conv = tf.nn.conv2d(state, art_W, strides=[1, 1, 1, 1], padding="VALID")
|
||||
relu = tf.nn.relu(tf.nn.bias_add(conv, art_b))
|
||||
|
||||
pooled = tf.reduce_max(relu, axis=1, keep_dims=True)
|
||||
art_pooled_outputs.append(pooled)
|
||||
|
||||
art_num_filters_total = self.art_num_filters * len(self.art_filter_sizes)
|
||||
|
||||
# [batch_size, art_num_filters]
|
||||
self.art_pool = tf.reshape(tf.concat(art_pooled_outputs, 3), [shape[0], 1, art_num_filters_total])
|
||||
|
||||
# [batch_size, sentences, num_filters]
|
||||
self.local = tf.reshape(state, [shape[0], -1, 2 * self.hidden_size])
|
||||
|
||||
# context vector for each sentence
|
||||
# [batch_size, sentences, art_num_filters]
|
||||
self.context = tf.tile(self.art_pool, [1, shape[1], 1])
|
||||
|
||||
# [batch_size, num_filters + art_num_filters]
|
||||
self.sentence = tf.concat([self.local, self.context], 2)
|
||||
|
||||
self.drop = tf.reshape(self.sentence, [shape[0], shape[1] , 2 * self.hidden_size + art_num_filters_total])
|
||||
self.drop = tf.nn.dropout(self.drop, keep_prob=self.keep_ratio)
|
||||
|
||||
self.fc_drop = fully_connected(self.drop, 1, activation_fn=tf.sigmoid)
|
||||
|
||||
a = tf.reshape(self.fc_drop, [shape[0], shape[1]])
|
||||
b = tf.cast(2, tf.int32)
|
||||
|
||||
#[batch_size, 2]
|
||||
self.highests = tf.nn.top_k(a, b)
|
||||
|
||||
self.logits1 = tf.reduce_mean(self.highests.values, axis=1)
|
||||
self.best_sentences = self.highests.indices
|
||||
self.logits0 = tf.ones_like(self.logits1) - self.logits1
|
||||
self.logits = tf.concat([tf.expand_dims(self.logits0, 1), tf.expand_dims(self.logits1, 1)], 1)
|
||||
|
||||
self.softmax = tf.nn.softmax(self.logits)
|
||||
self.xentropy = tf.nn.sparse_softmax_cross_entropy_with_logits(labels=self.output,
|
||||
logits=self.logits)
|
||||
self.loss = tf.reduce_mean(self.xentropy)
|
||||
self.predicted_label = tf.argmax(self.logits, 1)
|
||||
|
||||
self.accuracy = tf.reduce_mean(
|
||||
tf.cast(tf.equal(self.predicted_label, self.output), tf.float32))
|
||||
self.training_op = tf.train.AdamOptimizer(learning_rate=self.learning_rate).minimize(self.loss)
|
||||
|
||||
|
||||
def predict(self, unlabeled_batches, active_learning=False):
|
||||
hate_pred, indices_pred, hate_prob = list(), list(), list()
|
||||
saver = tf.train.Saver()
|
||||
with tf.Session() as self.sess:
|
||||
saver.restore(self.sess, "model/" + self.dataset + "/MICNN/micnn_model_2.ckpt")
|
||||
for i in range(len(unlabeled_batches) // 5000 + 1):
|
||||
print("Gathering labels for 5000 datapoints, batch #", i)
|
||||
sub = unlabeled_batches[i * 5000: min((i + 1) * 5000, len(unlabeled_batches))]
|
||||
batches = BatchIt(sub, self.batch_size, self.vocabs, True)
|
||||
for batch in batches:
|
||||
feed_dict = {self.train_inputs: np.array([b["article"] for b in batch]),
|
||||
self.sequence_length: np.array([l for b in batch for l in b["lengths"]]),
|
||||
self.keep_prob: 1,
|
||||
self.article_lens: np.array([b["sent_lengths"] for b in batch])
|
||||
}
|
||||
if self.pretrain:
|
||||
feed_dict[self.embedding_placeholder] = self.my_embeddings
|
||||
hate, indices, prob = self.sess.run([self.predicted_label, self.best_sentences, self.softmax],
|
||||
feed_dict=feed_dict)
|
||||
indices_pred.extend(list(indices))
|
||||
hate_pred.extend(list(hate))
|
||||
hate_prob.extend(list(prob))
|
||||
|
||||
if active_learning:
|
||||
return hate_pred, indices_pred, hate_prob
|
||||
else:
|
||||
return hate_pred, indices_pred
|
||||
|
||||
def get_feed_dict(self, batch, train=True):
|
||||
feed_dict = {self.train_inputs: np.array([b["article"] for b in batch]),
|
||||
self.sequence_length: np.array([l for b in batch for l in b["lengths"]]),
|
||||
self.keep_prob: self.keep_ratio if train else 1,
|
||||
self.output: np.array([b["labels"] for b in batch]),
|
||||
self.article_lens: np.array([b["sent_lengths"] for b in batch])
|
||||
}
|
||||
if self.pretrain:
|
||||
feed_dict[self.embedding_placeholder] = self.my_embeddings
|
||||
return feed_dict
|
||||
|
||||
def run_model(self, batches, dev_batches, test_batches):
|
||||
init = tf.global_variables_initializer()
|
||||
saver = tf.train.Saver()
|
||||
with tf.Session() as self.sess:
|
||||
# init.run()
|
||||
self.sess.run(init)
|
||||
epoch = 1
|
||||
while True:
|
||||
## Train
|
||||
epoch_loss, train_accuracy, test_accuracy = 0, 0, 0
|
||||
for batch in batches:
|
||||
feed_dict = self.get_feed_dict(batch)
|
||||
loss_val, _, train_val = self.sess.run([self.loss, self.training_op, self.accuracy],
|
||||
feed_dict=feed_dict)
|
||||
train_accuracy += train_val
|
||||
epoch_loss += loss_val
|
||||
## Dev
|
||||
for batch in dev_batches:
|
||||
feed_dict = self.get_feed_dict(batch, False)
|
||||
test_accuracy += self.accuracy.eval(feed_dict=feed_dict)
|
||||
|
||||
print(epoch, "Train accuracy:", train_accuracy / len(batches),
|
||||
"Loss:", epoch_loss / float(len(batches)),
|
||||
"Dev accuracy:", test_accuracy / len(dev_batches))
|
||||
epoch += 1
|
||||
if epoch == self.epochs:
|
||||
save_path = saver.save(self.sess, "model/" + self.dataset + "/MICNN/micnn_model_2.ckpt")
|
||||
break
|
||||
test_accuracy = 0
|
||||
hate_pred, hate_true = list(), list()
|
||||
for batch in test_batches:
|
||||
test_val, hate = self.sess.run([self.accuracy, self.predicted_label],
|
||||
feed_dict=self.get_feed_dict(batch, False))
|
||||
test_accuracy += test_val
|
||||
hate_pred.extend(list(hate))
|
||||
hate_true.extend([b["labels"] for b in batch])
|
||||
print("Test accuracy:", test_accuracy / len(test_batches),
|
||||
"F1:", f1_score(hate_true, hate_pred, average="binary"),
|
||||
"Precision", precision_score(hate_true, hate_pred),
|
||||
"Recall", recall_score(hate_true, hate_pred))
|
||||
|
||||
return self.get_sentences(batches, dev_batches, test_batches)
|
||||
|
||||
|
||||
def get_sentences(self, train_batches, dev_batches, test_batches):
|
||||
saver = tf.train.Saver()
|
||||
with tf.Session() as self.sess:
|
||||
saver.restore(self.sess, "model/" + self.dataset + "/MICNN/micnn_model_2.ckpt")
|
||||
train_sent, dev_sent, test_sent = list(), list(), list()
|
||||
for batch in train_batches:
|
||||
feed_dict = self.get_feed_dict(batch)
|
||||
train_sent.extend(list(self.sess.run(self.best_sentences, feed_dict=feed_dict)))
|
||||
|
||||
for batch in dev_batches:
|
||||
feed_dict = self.get_feed_dict(batch, False)
|
||||
dev_sent.extend(list(self.sess.run(self.best_sentences, feed_dict=feed_dict)))
|
||||
|
||||
for batch in test_batches:
|
||||
feed_dict = self.get_feed_dict(batch, False)
|
||||
test_sent.extend(list(self.sess.run(self.best_sentences, feed_dict=feed_dict)))
|
||||
|
||||
return train_sent, dev_sent, test_sent
|
||||
Loading…
Reference in New Issue
Block a user