def forward(self, trainer):
w_h = melt.init_weights([trainer.num_features, FLAGS.hidden_size
]) # create symbolic variables
w_o = melt.init_weights([FLAGS.hidden_size, 1])
py_x = self.model(trainer.X, w_h, w_o)
return py_x
def forward(self, trainer):
w_h = melt.init_weights([trainer.num_features, self.hidden_size], name = 'w_h') # create symbolic variables
b_h = melt.init_bias([1], name = 'b_h')
w_o = melt.init_weights([self.hidden_size, 1], name = 'w_o')
#self.weight = w_o #if add this can not cpickle dump
b_o = melt.init_bias([1], name = 'b_o')
py_x = self.model(trainer.X, w_h, b_h, w_o, b_o)
return py_x
def forward(self, trainer, FLAGS, numClass):
w_h1 = melt.init_weights([trainer.num_features, FLAGS.hidden_size
]) # create symbolic variables
w_h2 = melt.init_weights([FLAGS.hidden_size, FLAGS.hidden_size
]) # create symbolic variables
w_o = melt.init_weights([FLAGS.hidden_size, numClass])
py_x = self.model(trainer.X, w_h1, w_h2, w_o)
return py_x
def forward(self, trainer, FLAGS, numClass, gpu):
w_h = melt.init_weights([trainer.num_features, FLAGS.hidden_size
]) # create symbolic variables
b_h = melt.init_weights([FLAGS.hidden_size
]) # create symbolic variables
w_o = melt.init_weights([FLAGS.hidden_size, numClass])
b_o = melt.init_weights([numClass])
py_x = self.model(trainer.X, w_h, b_h, w_o, b_o, gpu)
return py_x
def forward(self, trainer):
opts = self.options
init_width = 0.5 / opts.emb_dim
vocab_size = trainer.num_features
emb = tf.Variable(
tf.random_uniform(
[vocab_size, opts.emb_dim], -init_width, init_width),
name="emb")
w_o = melt.init_weights([opts.emb_dim, 1], name = 'w_o') # create symbolic variables
b_o = melt.init_bias([1], name = 'b_o')
text_emb = tf.nn.embedding_lookup_sparse(emb, trainer.sp_ids, sp_weights = None, name = 'text_emb')
#return tf.matmul(self.activation(text_emb), w_o) + b_o
return tf.matmul(text_emb, w_o) + b_o
def forward(self, trainer):
w = melt.init_weights([trainer.num_features, 1], name = 'w')
b = melt.init_bias([1], name = 'b')
py_x = self.model(trainer.X, w, b)
return py_x
def forward(self, trainer):
w = melt.init_weights([trainer.num_features, 1])
py_x = self.model(trainer.X, w)
return py_x
def forward(self, trainer): w_h = melt.init_weights([trainer.num_features, FLAGS.hidden_size]) # create symbolic variables w_o = melt.init_weights([FLAGS.hidden_size, 1]) py_x = self.model(trainer.X, w_h, w_o) return py_x
def forward(self, trainer): w = melt.init_weights([trainer.num_features, 1]) py_x = self.model(trainer.X, w) return py_x
assert (trainset.num_features == testset.num_features)
num_features = trainset.num_features
print 'num_features: ', num_features
print 'trainSet size: ', trainset.num_instances()
print 'testSet size: ', testset.num_instances()
print 'batch_size:', batch_size, ' learning_rate:', learning_rate, ' num_epochs:', num_epochs
trainer = melt.gen_binary_classification_trainer(trainset)
#---------------- logistic regression
def model(X, w):
return melt.matmul(X, w)
w = melt.init_weights([num_features, 1])
py_x = model(trainer.X, w)
cost = tf.reduce_sum(tf.nn.sigmoid_cross_entropy_with_logits(py_x, trainer.Y))
train_op = tf.train.GradientDescentOptimizer(learning_rate).minimize(
cost) # construct optimizer
predict_op = tf.nn.sigmoid(py_x)
sess = tf.Session()
init = tf.initialize_all_variables()
sess.run(init)
teX, teY = testset.full_batch()
num_train_instances = trainset.num_instances()
for i in range(num_epochs):
predicts, cost_ = sess.run([predict_op, cost],
print "finish loading test set ", testset_file assert(trainset.num_features == testset.num_features) num_features = trainset.num_features print 'num_features: ', num_features print 'trainSet size: ', trainset.num_instances() print 'testSet size: ', testset.num_instances() print 'batch_size:', batch_size, ' learning_rate:', learning_rate, ' num_epochs:', num_epochs trainer = melt.gen_binary_classification_trainer(trainset) #---------------- logistic regression def model(X, w): return melt.matmul(X,w) w = melt.init_weights([num_features, 1]) py_x = model(trainer.X, w) cost = tf.reduce_sum(tf.nn.sigmoid_cross_entropy_with_logits(py_x, trainer.Y)) train_op = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost) # construct optimizer predict_op = tf.nn.sigmoid(py_x) sess = tf.Session() init = tf.initialize_all_variables() sess.run(init) teX, teY = testset.full_batch() num_train_instances = trainset.num_instances() for i in range(num_epochs): predicts, cost_ = sess.run([predict_op, cost], feed_dict = trainer.gen_feed_dict(teX, teY)) print i, 'auc:', roc_auc_score(teY, predicts), 'cost:', cost_