def ensemble_preds(dataset, nb_teachers, stdnt_data):
"""
Given a dataset, a number of teachers, and some input data, this helper
function queries each teacher for predictions on the data and returns
all predictions in a single array. (That can then be aggregated into
one single prediction per input using aggregation.py (cf. function
prepare_student_data() below)
:param dataset: string corresponding to mnist, cifar10, or svhn
:param nb_teachers: number of teachers (in the ensemble) to learn from
:param stdnt_data: unlabeled student training data
:return: 3d array (teacher id, sample id, probability per class)
"""
# Compute shape of array that will hold probabilities produced by each
# teacher, for each training point, and each output class
result_shape = (nb_teachers, len(stdnt_data), FLAGS.nb_labels)
# Create array that will hold result
result = np.zeros(result_shape, dtype=np.float32)
# Get predictions from each teacher
for teacher_id in xrange(nb_teachers):
# Compute path of checkpoint file for teacher model with ID teacher_id
if FLAGS.deeper:
ckpt_path = FLAGS.teachers_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_teachers_' + str(teacher_id) + '_deep.ckpt-' + str(FLAGS.teachers_max_steps - 1) #NOLINT(long-line)
else:
ckpt_path = FLAGS.teachers_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_teachers_' + str(teacher_id) + '.ckpt-' + str(FLAGS.teachers_max_steps - 1) # NOLINT(long-line)
# Get predictions on our training data and store in result array
result[teacher_id] = deep_cnn.softmax_preds(stdnt_data, ckpt_path)
# This can take a while when there are a lot of teachers so output status
print("Computed Teacher " + str(teacher_id) + " softmax predictions")
return result
def ensemble_preds(dataset, nb_teachers, stdnt_data):
"""
Given a dataset, a number of teachers, and some input data, this helper
function queries each teacher for predictions on the data and returns
all predictions in a single array. (That can then be aggregated into
one single prediction per input using aggregation.py (cf. function
prepare_student_data() below)
:param dataset: string corresponding to mnist, cifar10, or svhn
:param nb_teachers: number of teachers (in the ensemble) to learn from
:param stdnt_data: unlabeled student training data
:return: 3d array (teacher id, sample id, probability per class)
"""
# Compute shape of array that will hold probabilities produced by each
# teacher, for each training point, and each output class
result_shape = (nb_teachers, len(stdnt_data), FLAGS.nb_labels)
# Create array that will hold result
result = np.zeros(result_shape, dtype=np.float32)
# Get predictions from each teacher
for teacher_id in xrange(nb_teachers):
# Compute path of checkpoint file for teacher model with ID teacher_id
if FLAGS.deeper:
ckpt_path = FLAGS.teachers_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_teachers_' + str(teacher_id) + '_deep.ckpt-' + str(FLAGS.teachers_max_steps - 1) #NOLINT(long-line)
else:
ckpt_path = FLAGS.teachers_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_teachers_' + str(teacher_id) + '.ckpt-' + str(FLAGS.teachers_max_steps - 1) # NOLINT(long-line)
# Get predictions on our training data and store in result array
result[teacher_id] = deep_cnn.softmax_preds(stdnt_data, ckpt_path)
# This can take a while when there are a lot of teachers so output status
print("Computed Teacher " + str(teacher_id) + " softmax predictions")
return result
def train_teacher(dataset, nb_teachers, teacher_id):
"""
This function trains a teacher (teacher id) among an ensemble of nb_teachers
models for the dataset specified.
:param dataset: string corresponding to dataset (svhn, cifar10)
:param nb_teachers: total number of teachers in the ensemble
:param teacher_id: id of the teacher being trained
:return: True if everything went well
"""
# If working directories do not exist, create them
assert input.create_dir_if_needed(FLAGS.data_dir)
assert input.create_dir_if_needed(FLAGS.train_dir)
# Load the dataset
if dataset == 'svhn':
train_data, train_labels, test_data, test_labels = input.ld_svhn(
extended=True)
elif dataset == 'cifar10':
train_data, train_labels, test_data, test_labels = input.ld_cifar10()
elif dataset == 'mnist':
train_data, train_labels, test_data, test_labels = input.ld_mnist()
else:
print("Check value of dataset flag")
return False
if FLAGS.teacher_data_share:
train_data = train_data[:FLAGS.teacher_data_share]
train_labels = train_labels[:FLAGS.teacher_data_share]
# Retrieve subset of data for this teacher
data, labels = input.partition_dataset(train_data, train_labels,
nb_teachers, teacher_id)
print("Length of training data: " + str(len(labels)))
# Define teacher checkpoint filename and full path
if FLAGS.deeper:
filename = str(nb_teachers) + '_teachers_' + str(
teacher_id) + '_deep.ckpt'
else:
filename = str(nb_teachers) + '_teachers_' + str(teacher_id) + '.ckpt'
ckpt_path = FLAGS.train_dir + '/' + str(dataset) + '_' + filename
# Perform teacher training
assert deep_cnn.train(data, labels, ckpt_path)
# Append final step value to checkpoint for evaluation
ckpt_path_final = ckpt_path + '-' + str(FLAGS.max_steps - 1)
# Retrieve teacher probability estimates on the test data
teacher_preds = deep_cnn.softmax_preds(test_data, ckpt_path_final)
# Compute teacher accuracy
precision = metrics.accuracy(teacher_preds, test_labels)
print('Precision of teacher after training: ' + str(precision))
return True
def train_teacher(dataset, nb_teachers, teacher_id):
"""
This function trains a teacher (teacher id) among an ensemble of nb_teachers
models for the dataset specified.
:param dataset: string corresponding to dataset (svhn, cifar10)
:param nb_teachers: total number of teachers in the ensemble
:param teacher_id: id of the teacher being trained
:return: True if everything went well
"""
# If working directories do not exist, create them
assert input.create_dir_if_needed(FLAGS.data_dir)
assert input.create_dir_if_needed(FLAGS.train_dir)
# Load the dataset
if dataset == 'svhn':
train_data,train_labels,test_data,test_labels = input.ld_svhn(extended=True)
elif dataset == 'cifar10':
train_data, train_labels, test_data, test_labels = input.ld_cifar10()
elif dataset == 'mnist':
train_data, train_labels, test_data, test_labels = input.ld_mnist()
else:
print("Check value of dataset flag")
return False
# Retrieve subset of data for this teacher
data, labels = input.partition_dataset(train_data,
train_labels,
nb_teachers,
teacher_id)
print("Length of training data: " + str(len(labels)))
# Define teacher checkpoint filename and full path
if FLAGS.deeper:
filename = str(nb_teachers) + '_teachers_' + str(teacher_id) + '_deep.ckpt'
else:
filename = str(nb_teachers) + '_teachers_' + str(teacher_id) + '.ckpt'
ckpt_path = FLAGS.train_dir + '/' + str(dataset) + '_' + filename
# Perform teacher training
assert deep_cnn.train(data, labels, ckpt_path)
# Append final step value to checkpoint for evaluation
ckpt_path_final = ckpt_path + '-' + str(FLAGS.max_steps - 1)
# Retrieve teacher probability estimates on the test data
teacher_preds = deep_cnn.softmax_preds(test_data, ckpt_path_final)
# Compute teacher accuracy
precision = metrics.accuracy(teacher_preds, test_labels)
print('Precision of teacher after training: ' + str(precision))
return True
def train_student(dataset, nb_teachers):
"""
This function trains a student using predictions made by an ensemble of
teachers. The student and teacher models are trained using the same
neural network architecture.
:param dataset: string corresponding to mnist, cifar10, or svhn
:param nb_teachers: number of teachers (in the ensemble) to learn from
:return: True if student training went well
"""
assert input.create_dir_if_needed(FLAGS.train_dir)
# Call helper function to prepare student data using teacher predictions
stdnt_dataset = prepare_student_data(dataset, nb_teachers, save=True)
# Unpack the student dataset
stdnt_data, stdnt_labels, stdnt_test_data, stdnt_test_labels = stdnt_dataset
# Prepare checkpoint filename and path
if FLAGS.deeper:
ckpt_path = FLAGS.train_dir + '/' + str(dataset) + '_' + str(
nb_teachers) + '_student_deeper.ckpt' #NOLINT(long-line)
else:
ckpt_path = FLAGS.train_dir + '/' + str(dataset) + '_' + str(
nb_teachers) + '_student.ckpt' # NOLINT(long-line)
# Start student training
assert deep_cnn.train(stdnt_data, stdnt_labels, ckpt_path)
# Compute final checkpoint name for student (with max number of steps)
ckpt_path_final = ckpt_path + '-' + str(FLAGS.max_steps - 1)
# Compute student label predictions on remaining chunk of test set
student_preds = deep_cnn.softmax_preds(stdnt_test_data, ckpt_path_final)
# Compute teacher accuracy
precision = metrics.accuracy(student_preds, stdnt_test_labels)
print('Precision of student after training: ' + str(precision))
return True
def ensemble_preds(dataset, nb_teachers, stdnt_data):
"""
Given a dataset, a number of teachers, and some input data, this helper
function queries each teacher for predictions on the data and returns
all predictions in a single array. (That can then be aggregated into
one single prediction per input using aggregation.py (cf. function
prepare_student_data() below)
:param dataset: string corresponding to mnist, cifar10, or svhn
:param nb_teachers: number of teachers (in the ensemble) to learn from
:param stdnt_data: unlabeled student training data#没有标记的学生的训练数据
:return: 3d array (teacher id, sample id, probability per class) #3d数组(教师id、样本id、每个类的概率)
"""
#计算每个教师，每个训练点和每个输出类所产生概率的数组形式
# Compute shape of array that will hold probabilities produced by each
# teacher, for each training point, and each output class
result_shape = (nb_teachers, len(stdnt_data), FLAGS.nb_labels)#stdnt_data:没有标记的学生的训练数据
# nb_labels:输出类的数量
# Create array that will hold result创建将保存结果的数组
result = np.zeros(result_shape, dtype=np.float32) #numpy.zeros返回给定形状和类型的新数组，用零填充。
# Get predictions from each teacher 从每个老师那里得到预测
for teacher_id in xrange(nb_teachers):
# Compute path of checkpoint file for teacher model with ID teacher_id 为ID为teacher_id的教师模型计算检查点文件的路径
if FLAGS.deeper:
ckpt_path = FLAGS.teachers_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_teachers_' + str(teacher_id) + '_deep.ckpt-' + str(FLAGS.teachers_max_steps - 1) #NOLINT(long-line)
else:
ckpt_path = FLAGS.teachers_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_teachers_' + str(teacher_id) + '.ckpt-' + str(FLAGS.teachers_max_steps - 1) # NOLINT(long-line)
# Get predictions on our training data and store in result array 对我们的训练数据进行预测并存储在结果数组中
result[teacher_id] = deep_cnn.softmax_preds(stdnt_data, ckpt_path)
# This can take a while when there are a lot of teachers so output status
print("Computed Teacher " + str(teacher_id) + " softmax predictions")
return result
def train_student(dataset, nb_teachers):
"""
This function trains a student using predictions made by an ensemble of
teachers. The student and teacher models are trained using the same
neural network architecture.
:param dataset: string corresponding to mnist, cifar10, or svhn 与mnist、cifar10或svhn相对应的字符串
:param nb_teachers: number of teachers (in the ensemble) to learn from
:return: True if student training went well
"""
assert input.create_dir_if_needed(FLAGS.train_dir)
# Call helper function to prepare student data using teacher predictions调用助手函数，使用教师预测来准备学生数据
stdnt_dataset = prepare_student_data(dataset, nb_teachers, save=True)
# Unpack the student dataset 打开学生的数据集
stdnt_data, stdnt_labels, stdnt_test_data, stdnt_test_labels = stdnt_dataset
# Prepare checkpoint filename and path 准备检查点文件名和路径
if FLAGS.deeper:
ckpt_path = FLAGS.train_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_student_deeper.ckpt' #NOLINT(long-line)
else:
ckpt_path = FLAGS.train_dir + '/' + str(dataset) + '_' + str(nb_teachers) + '_student.ckpt' # NOLINT(long-line)
# Start student training
assert deep_cnn.train(stdnt_data, stdnt_labels, ckpt_path)
# Compute final checkpoint name for student (with max number of steps) 计算学生的最终检查点名称（最大步数）
ckpt_path_final = ckpt_path + '-' + str(FLAGS.max_steps - 1)
# Compute student label predictions on remaining chunk of test set 在剩余的测试集上计算学生标签预测
student_preds = deep_cnn.softmax_preds(stdnt_test_data, ckpt_path_final)
# Compute teacher accuracy
precision = metrics.accuracy(student_preds, stdnt_test_labels)
print('Precision of student after training: ' + str(precision))
return True
