# ---------------------------------------------------------------#
# astroNN.nn.losses: losses
# ---------------------------------------------------------------#
import tensorflow as tf
try: # remove once TF2 is finalized
from tensorflow import distributions as tfd
except ImportError:
import tensorflow_probability as tfp
tfd = tfp.distributions
from astroNN.config import MAGIC_NUMBER
from astroNN.config import keras_import_manager
from astroNN.nn import magic_correction_term, nn_obj_lookup
keras = keras_import_manager()
epsilon = keras.backend.epsilon
Model = keras.models.Model
[docs]def mean_squared_error(y_true, y_pred):
"""
Calculate mean square error losses
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Mean Squared Error
:rtype: tf.Tensor
:History: 2017-Nov-16 - Written - Henry Leung (University of Toronto)
"""
return tf.reduce_mean(tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true),
tf.square(y_true - y_pred)), axis=-1) * magic_correction_term(y_true)
[docs]def mse_lin_wrapper(var, labels_err):
"""
Calculate predictive variance, and takes account of labels error in Bayesian Neural Network
:param var: Predictive Variance
:type var: Union(tf.Tensor, tf.Variable)
:param labels_err: Known labels error, give zeros if unknown/unavailable
:type labels_err: Union(tf.Tensor, tf.Variable)
:return: Robust MSE function for labels prediction neurones, which matches Keras losses API
:rtype: function
:Returned Funtion Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Prediction
| Return (*tf.Tensor*): Robust Mean Squared Error
:History: 2017-Nov-16 - Written - Henry Leung (University of Toronto)
"""
def mse_lin(y_true, y_pred):
return robust_mse(y_true, y_pred, var, labels_err)
mse_lin.__name__ = 'mse_lin_wrapper' # set the name to be the same as parent so it can be found
return mse_lin
[docs]def mse_var_wrapper(lin, labels_err):
"""
Calculate predictive variance, and takes account of labels error in Bayesian Neural Network
:param lin: Prediction
:type lin: Union(tf.Tensor, tf.Variable)
:param labels_err: Known labels error, give zeros if unknown/unavailable
:type labels_err: Union(tf.Tensor, tf.Variable)
:return: Robust MSE function for predictive variance neurones which matches Keras losses API
:rtype: function
:Returned Funtion Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Predictive Variance
| Return (*tf.Tensor*): Robust Mean Squared Error
:History: 2017-Nov-16 - Written - Henry Leung (University of Toronto)
"""
def mse_var(y_true, y_pred):
return robust_mse(y_true, lin, y_pred, labels_err)
mse_var.__name__ = 'mse_var_wrapper' # set the name to be the same as parent so it can be found
return mse_var
[docs]def robust_mse(y_true, y_pred, variance, labels_err):
"""
Calculate predictive variance, and takes account of labels error in Bayesian Neural Network
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:param variance: Log Predictive Variance
:type variance: Union(tf.Tensor, tf.Variable)
:param labels_err: Known labels error, give zeros if unknown/unavailable
:type labels_err: Union(tf.Tensor, tf.Variable)
:return: Robust Mean Squared Error, can be used directly with Tensorflow
:rtype: tf.Tensor
:History: 2018-April-07 - Written - Henry Leung (University of Toronto)
"""
# labels_err still contains magic_number
labels_err_y = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), labels_err)
# Neural Net is predicting log(var), so take exp, takes account the target variance, and take log back
y_pred_corrected = tf.log(tf.exp(variance) + tf.square(labels_err_y))
wrapper_output = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true),
0.5 * tf.square(y_true - y_pred) * (tf.exp(-y_pred_corrected)) + 0.5 *
y_pred_corrected)
return tf.reduce_mean(wrapper_output, axis=-1) * magic_correction_term(y_true)
[docs]def mean_absolute_error(y_true, y_pred):
"""
Calculate mean absolute error, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Mean Absolute Error
:rtype: tf.Tensor
:History: 2018-Jan-14 - Written - Henry Leung (University of Toronto)
"""
return tf.reduce_mean(tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true),
tf.abs(y_true - y_pred)), axis=-1) * magic_correction_term(y_true)
[docs]def mean_absolute_percentage_error(y_true, y_pred):
"""
Calculate mean absolute percentage error, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Mean Absolute Percentage Error
:rtype: tf.Tensor
:History: 2018-Feb-17 - Written - Henry Leung (University of Toronto)
"""
tf_inf = tf.cast(tf.constant(1) / tf.constant(0), tf.float32)
epsilon_tensor = tf.cast(tf.constant(keras.backend.epsilon()), tf.float32)
diff = tf.abs((y_true - y_pred) / tf.clip_by_value(tf.abs(y_true), epsilon_tensor, tf_inf))
diff_corrected = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), diff)
return 100. * tf.reduce_mean(diff_corrected, axis=-1) * magic_correction_term(y_true)
[docs]def mean_squared_logarithmic_error(y_true, y_pred):
"""
Calculate mean squared logarithmic error, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Mean Squared Logarithmic Error
:rtype: tf.Tensor
:History: 2018-Feb-17 - Written - Henry Leung (University of Toronto)
"""
tf_inf = tf.cast(tf.constant(1) / tf.constant(0), tf.float32)
epsilon_tensor = tf.cast(tf.constant(keras.backend.epsilon()), tf.float32)
first_log = tf.log(tf.clip_by_value(y_pred, epsilon_tensor, tf_inf) + 1.)
second_log = tf.log(tf.clip_by_value(y_true, epsilon_tensor, tf_inf) + 1.)
log_diff = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), tf.square(first_log - second_log))
return tf.reduce_mean(log_diff, axis=-1) * magic_correction_term(y_true)
[docs]def mean_error(y_true, y_pred):
"""
Calculate mean error as a way to get the bias in prediction, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Mean Error
:rtype: tf.Tensor
:History: 2018-May-22 - Written - Henry Leung (University of Toronto)
"""
return tf.reduce_mean(tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), y_true - y_pred),
axis=-1) * magic_correction_term(y_true)
[docs]def mean_percentage_error(y_true, y_pred):
"""
Calculate mean percentage error, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Mean Percentage Error
:rtype: tf.Tensor
:History: 2018-Jun-06 - Written - Henry Leung (University of Toronto)
"""
tf_inf = tf.cast(tf.constant(1) / tf.constant(0), tf.float32)
epsilon_tensor = tf.cast(tf.constant(keras.backend.epsilon()), tf.float32)
diff = y_true - y_pred / tf.clip_by_value(y_true, epsilon_tensor, tf_inf)
diff_corrected = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), diff)
return 100. * tf.reduce_mean(diff_corrected, axis=-1) * magic_correction_term(y_true)
[docs]def categorical_crossentropy(y_true, y_pred, from_logits=False):
"""
Categorical cross-entropy between an output tensor and a target tensor, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:param from_logits: From logits space or not. If you want to use logits, please use from_logits=True
:type from_logits: boolean
:return: Categorical Cross-Entropy
:rtype: tf.Tensor
:History: 2018-Jan-14 - Written - Henry Leung (University of Toronto)
"""
# calculate correction term first
correction = magic_correction_term(y_true)
# Deal with magic number
y_true = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), y_true)
# Note: tf.nn.softmax_cross_entropy_with_logits_v2 expects logits, we expects probabilities by default.
if not from_logits:
epsilon_tensor = tf.cast(tf.constant(keras.backend.epsilon()), tf.float32)
# scale preds so that the class probas of each sample sum to 1
y_pred /= tf.reduce_sum(y_pred, len(y_pred.get_shape()) - 1, True)
# manual computation of crossentropy
y_pred = tf.clip_by_value(y_pred, epsilon_tensor, 1. - epsilon_tensor)
return - tf.reduce_sum(y_true * tf.log(y_pred), len(y_pred.get_shape()) - 1) * correction
else:
return tf.nn.softmax_cross_entropy_with_logits_v2(labels=y_true, logits=y_pred) * correction
[docs]def binary_crossentropy(y_true, y_pred, from_logits=False):
"""
Binary cross-entropy between an output tensor and a target tensor, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:param from_logits: From logits space or not. If you want to use logits, please use from_logits=True
:type from_logits: boolean
:return: Binary Cross-Entropy
:rtype: tf.Tensor
:History: 2018-Jan-14 - Written - Henry Leung (University of Toronto)
"""
# Note: tf.nn.sigmoid_cross_entropy_with_logits expects logits, we expects probabilities by default.
if not from_logits:
epsilon_tensor = tf.cast(tf.constant(keras.backend.epsilon()), tf.float32)
# transform back to logits
y_pred = tf.clip_by_value(y_pred, epsilon_tensor, 1. - epsilon_tensor)
y_pred = tf.log(y_pred / (1. - y_pred))
cross_entropy = tf.nn.sigmoid_cross_entropy_with_logits(labels=y_true, logits=y_pred)
corrected_cross_entropy = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(cross_entropy), cross_entropy)
return tf.reduce_mean(corrected_cross_entropy, axis=-1) * magic_correction_term(y_true)
[docs]def bayesian_categorical_crossentropy_wrapper(logit_var):
"""
| Categorical crossentropy between an output tensor and a target tensor for Bayesian Neural Network
| equation (12) of arxiv:1703.04977
:param logit_var: Predictive variance
:type logit_var: Union(tf.Tensor, tf.Variable)
:return: Robust categorical_crossentropy function for predictive variance neurones which matches Keras losses API
:rtype: function
:Returned Function Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Prediction in logits space
| Return (*tf.Tensor*): Robust categorical crossentropy
:History: 2018-Mar-15 - Written - Henry Leung (University of Toronto)
"""
# y_pred is logits
def bayesian_crossentropy(y_true, y_pred):
return robust_categorical_crossentropy(y_true, y_pred, logit_var)
# set the name to be the same as parent so it can be found
bayesian_crossentropy.__name__ = 'bayesian_categorical_crossentropy_wrapper'
return bayesian_crossentropy
[docs]def bayesian_categorical_crossentropy_var_wrapper(logits):
"""
| Categorical crossentropy between an output tensor and a target tensor for Bayesian Neural Network
| equation (12) of arxiv:1703.04977
:param logits: Prediction in logits space
:type logits: Union(tf.Tensor, tf.Variable)
:return: Robust categorical_crossentropy function for predictive variance neurones which matches Keras losses API
:rtype: function
:Returned Function Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Predictive variance in logits space
| Return (*tf.Tensor*): Robust categorical crossentropy
:History: 2018-Mar-15 - Written - Henry Leung (University of Toronto)
"""
# y_pred is predictive entropy
def bayesian_crossentropy(y_true, y_pred):
return robust_categorical_crossentropy(y_true, logits, y_pred)
# set the name to be the same as parent so it can be found
bayesian_crossentropy.__name__ = 'bayesian_categorical_crossentropy_var_wrapper'
return bayesian_crossentropy
[docs]def robust_categorical_crossentropy(y_true, y_pred, logit_var):
"""
Calculate categorical accuracy, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction in logits space
:type y_pred: Union(tf.Tensor, tf.Variable)
:param logit_var: Predictive variance in logits space
:type logit_var: Union(tf.Tensor, tf.Variable)
:return: categorical cross-entropy
:rtype: tf.Tensor
:History: 2018-Mar-15 - Written - Henry Leung (University of Toronto)
"""
variance_depressor = tf.reduce_mean(tf.exp(logit_var) - tf.ones_like(logit_var))
undistorted_loss = categorical_crossentropy(y_true, y_pred, from_logits=True)
dist = tfd.Normal(loc=y_pred, scale=logit_var)
mc_result = tf.map_fn(
lambda x: -tf.nn.elu(undistorted_loss - categorical_crossentropy(y_true, x, from_logits=True)),
dist.sample([25]), dtype=tf.float32)
variance_loss = tf.reduce_mean(mc_result, axis=0) * undistorted_loss
return (variance_loss + undistorted_loss + variance_depressor) * magic_correction_term(y_true)
[docs]def bayesian_binary_crossentropy_wrapper(logit_var):
"""
| Binary crossentropy between an output tensor and a target tensor for Bayesian Neural Network
| equation (12) of arxiv:1703.04977
:param logit_var: Predictive variance
:type logit_var: Union(tf.Tensor, tf.Variable)
:return: Robust binary_crossentropy function for predictive variance neurones which matches Keras losses API
:rtype: function
:Returned Function Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Prediction in logits space
| Return (*tf.Tensor*): Robust binary crossentropy
:History: 2018-Mar-15 - Written - Henry Leung (University of Toronto)
"""
# y_pred is logits
def bayesian_crossentropy(y_true, y_pred):
return robust_binary_crossentropy(y_true, y_pred, logit_var)
# set the name to be the same as parent so it can be found
bayesian_crossentropy.__name__ = 'bayesian_binary_crossentropy_wrapper'
return bayesian_crossentropy
[docs]def bayesian_binary_crossentropy_var_wrapper(logits):
"""
| Binary crossentropy between an output tensor and a target tensor for Bayesian Neural Network
| equation (12) of arxiv:1703.04977
:param logits: Prediction in logits space
:type logits: Union(tf.Tensor, tf.Variable)
:return: Robust binary_crossentropy function for predictive variance neurones which matches Keras losses API
:rtype: function
:Returned Function Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Predictive variance in logits space
| Return (*tf.Tensor*): Robust binary crossentropy
:History: 2018-Mar-15 - Written - Henry Leung (University of Toronto)
"""
# y_pred is predictive entropy
def bayesian_crossentropy(y_true, y_pred):
return robust_binary_crossentropy(y_true, logits, y_pred)
# set the name to be the same as parent so it can be found
bayesian_crossentropy.__name__ = 'bayesian_binary_crossentropy_var_wrapper'
return bayesian_crossentropy
[docs]def robust_binary_crossentropy(y_true, y_pred, logit_var):
"""
Calculate binary accuracy, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction in logits space
:type y_pred: Union(tf.Tensor, tf.Variable)
:param logit_var: Predictive variance in logits space
:type logit_var: Union(tf.Tensor, tf.Variable)
:return: categorical cross-entropy
:rtype: tf.Tensor
:History: 2018-Mar-15 - Written - Henry Leung (University of Toronto)
"""
variance_depressor = tf.reduce_mean(tf.exp(logit_var) - tf.ones_like(logit_var))
undistorted_loss = binary_crossentropy(y_true, y_pred, from_logits=True)
dist = tfd.Normal(loc=y_pred, scale=logit_var)
mc_result = tf.map_fn(
lambda x: -tf.nn.elu(undistorted_loss - binary_crossentropy(y_true, x, from_logits=True)),
dist.sample([25]), dtype=tf.float32)
variance_loss = tf.reduce_mean(mc_result, axis=0) * undistorted_loss
return (variance_loss + undistorted_loss + variance_depressor) * magic_correction_term(y_true)
def nll(y_true, y_pred):
"""
Calculate negative log likelihood
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Negative log likelihood
:rtype: tf.Tensor
:History: 2018-Jan-30 - Written - Henry Leung (University of Toronto)
"""
# astroNN binary_cross_entropy gives the mean over the last axis. we require the sum
return tf.reduce_sum(binary_crossentropy(y_true, y_pred), axis=-1)
[docs]def categorical_accuracy(y_true, y_pred):
"""
Calculate categorical accuracy, ignoring the magic number
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Categorical Classification Accuracy
:rtype: tf.Tensor
:History: 2018-Jan-21 - Written - Henry Leung (University of Toronto)
"""
y_true = tf.where(tf.equal(y_true, MAGIC_NUMBER), tf.zeros_like(y_true), y_true)
return tf.cast(tf.equal(tf.argmax(y_true, axis=-1), tf.argmax(y_pred, axis=-1)),
tf.float32) * magic_correction_term(y_true)
[docs]def binary_accuracy(from_logits=False):
"""
Calculate binary accuracy, ignoring the magic number
:param from_logits: From logits space or not. If you want to use logits, please use from_logits=True
:type from_logits: boolean
:return: Function for Binary classification accuracy which matches Keras losses API
:rtype: function
:Returned Funtion Parameter:
| **function(y_true, y_pred)**
| - **y_true** (*tf.Tensor*): Ground Truth
| - **y_pred** (*tf.Tensor*): Prediction
| Return (*tf.Tensor*): Binary Classification Accuracy
:History: 2018-Jan-31 - Written - Henry Leung (University of Toronto)
"""
# DO NOT correct y_true for magic number, just let it goes wrong and then times a correction terms
def binary_accuracy_internal(y_true, y_pred):
if from_logits:
y_pred = tf.nn.sigmoid(y_pred)
return tf.reduce_mean(tf.cast(tf.equal(y_true, tf.round(y_pred)), tf.float32), axis=-1) * magic_correction_term(
y_true)
binary_accuracy_internal.__name__ = 'binary_accuracy' # set the name to be displayed in TF/Keras log
return binary_accuracy_internal
[docs]def zeros_loss(y_true, y_pred):
"""
Always return zeros
:param y_true: Ground Truth
:type y_true: Union(tf.Tensor, tf.Variable)
:param y_pred: Prediction
:type y_pred: Union(tf.Tensor, tf.Variable)
:return: Zeros
:rtype: tf.Tensor
:History: 2018-May-24 - Written - Henry Leung (University of Toronto)
"""
return tf.reduce_mean(0. * y_true + 0. * y_pred, axis=-1)
# Just alias functions
mse = mean_squared_error
mae = mean_absolute_error
mape = mean_absolute_percentage_error
msle = mean_squared_logarithmic_error
# legacy support
mse_lin = mse_lin_wrapper
mse_var = mse_var_wrapper
def losses_lookup(identifier):
"""
Lookup astroNN.nn.losses function by name
:param identifier: identifier
:type identifier: str
:return: Looked up function
:rtype: function
:History: 2018-Apr-28 - Written - Henry Leung (University of Toronto)
"""
return nn_obj_lookup(identifier, module_obj=globals(), module_name=__name__)
