def _train_model(self, train_set, train_ref, validation_set,
validation_ref):
""" Train the model.
:param train_set: training set
:param train_ref: training reference data
:param validation_set: validation set
:param validation_ref: validation reference data
:return: self
"""
shuff = zip(train_set, train_ref)
for i in range(self.finetune_num_epochs):
np.random.shuffle(shuff)
batches = [
_
for _ in utilities.gen_batches(shuff, self.finetune_batch_size)
]
for batch in batches:
x_batch, y_batch = zip(*batch)
self.tf_session.run(self.train_step,
feed_dict={
self.input_data: x_batch,
self.input_labels: y_batch,
self.keep_prob: self.dropout
})
if validation_set is not None:
self._run_validation_error_and_summaries(
i, validation_set, validation_ref)
def _train_model(self, train_set, train_labels, validation_set,
validation_labels):
""" Train the model.
:param train_set: training set
:param train_labels: training labels
:param validation_set: validation set
:param validation_labels: validation labels
:return: self
"""
for i in range(self.num_epochs):
shuff = list(zip(train_set, train_labels))
np.random.shuffle(shuff)
batches = [
_ for _ in utilities.gen_batches(shuff, self.batch_size)
]
for batch in batches:
x_batch, y_batch = zip(*batch)
self.tf_session.run(self.train_step,
feed_dict={
self.input_data: x_batch,
self.input_labels: y_batch
})
if validation_set is not None:
self._run_validation_error_and_summaries(
i, validation_set, validation_labels)
def _train_model(self,
train_set,
train_labels,
validation_set=None,
validation_labels=None):
""" Train the model.
:param train_set: training set
:param train_labels: training labels
:param validation_set: validation set
:param validation_labels: validation labels
:return:
"""
shuff = zip(train_set, train_labels)
for i in range(self.num_epochs):
np.random.shuffle(shuff)
batches = [
_ for _ in utilities.gen_batches(shuff, self.batch_size)
]
for batch in batches:
x_batch, y_batch = zip(*batch)
feed = self._create_feed(x_batch, y_batch, self.dropout)
self.tf_session.run(self.train_step, feed_dict=feed)
if validation_set is not None:
self._run_validation_error_and_summaries(
i, validation_set, validation_labels)
def _run_train_step(self, train_set):
""" Run a training step. A training step is made by randomly shuffling the training set,
divide into batches and run the variable update nodes for each batch.
:param train_set: training set
:return: self
"""
np.random.shuffle(train_set)
batches = [_ for _ in utilities.gen_batches(train_set, self.batch_size)]
updates = [self.w_upd8, self.bh_upd8, self.bv_upd8]
for batch in batches:
self.tf_session.run(updates, feed_dict=self._create_feed_dict(batch))
def _run_train_step(self, train_set):
""" Run a training step. A training step is made by randomly corrupting the training set,
randomly shuffling it, divide it into batches and run the optimizer for each batch.
:param train_set: training set
:return: self
"""
x_corrupted = self._corrupt_input(train_set)
shuff = zip(train_set, x_corrupted)
np.random.shuffle(shuff)
batches = [_ for _ in utilities.gen_batches(shuff, self.batch_size)]
for batch in batches:
x_batch, x_corr_batch = zip(*batch)
tr_feed = {self.input_data: x_batch, self.input_data_corr: x_corr_batch}
self.tf_session.run(self.train_step, feed_dict=tr_feed)
def _train_model(self, train_set, validation_set):
""" Train the model.
:param train_set: training set
:param validation_set: validation set
:return: self
"""
batches = [
_ for _ in utilities.gen_batches(train_set, self.batch_size)
]
for i in range(self.num_epochs):
for batch in batches:
feed = self._create_feed(batch, self.dropout)
self.tf_session.run(self.train_step, feed_dict=feed)
if i % 5 == 0:
if validation_set is not None:
self._run_validation_error_and_summaries(i, validation_set)