def update_state(self, y_true, y_pred, sample_weight=None):
"""Accumulates confusion matrix statistics.
Args:
y_true: The ground truth values.
y_pred: The predicted values.
sample_weight: Optional weighting of each example. Defaults to 1. Can be a
`Tensor` whose rank is either 0, or the same rank as `y_true`, and must
be broadcastable to `y_true`.
Returns:
Update op.
"""
deps = []
if not self._built:
y_true = tf.ensure_shape(y_true,
(self.num_thresholds, self.num_classes))
self._build(y_true.shape)
if self.multi_label or (self.label_weights is not None):
# y_true should have shape (number of examples, number of labels).
shapes = [(y_true, ('N', 'L'))]
if self.multi_label:
# TP, TN, FP, and FN should all have shape
# (number of thresholds, number of labels).
shapes.extend([(self.true_positives, ('T', 'L')),
(self.true_negatives, ('T', 'L')),
(self.false_positives, ('T', 'L')),
(self.false_negatives, ('T', 'L'))])
if self.label_weights is not None:
# label_weights should be of lenght equal to the number of labels.
shapes.append((self.label_weights, ('L', )))
deps = [
check_ops.assert_shapes(
shapes, message='Number of labels is not consistent.')
]
# Only forward label_weights to update_confusion_matrix_variables when
# multi_label is False. Otherwise the averaging of individual label AUCs is
# handled in AUC.result
label_weights = None if self.multi_label else self.label_weights
with ops.control_dependencies(deps):
metrics_utils.update_confusion_matrix_variables(
{
metrics_utils.ConfusionMatrix.TRUE_POSITIVES:
self.true_positives,
metrics_utils.ConfusionMatrix.TRUE_NEGATIVES:
self.true_negatives,
metrics_utils.ConfusionMatrix.FALSE_POSITIVES:
self.false_positives,
metrics_utils.ConfusionMatrix.FALSE_NEGATIVES:
self.false_negatives,
},
y_true,
y_pred,
self.thresholds,
sample_weight=sample_weight,
multi_label=self.multi_label,
label_weights=label_weights)
def update_state(self, y_true, y_pred, sample_weight=None):
metrics_utils.update_confusion_matrix_variables(
{
metrics_utils.ConfusionMatrix.TRUE_POSITIVES:
self.true_positives,
metrics_utils.ConfusionMatrix.FALSE_NEGATIVES:
self.false_negatives
},
y_true,
y_pred,
threshold=self.threshold,
)
def update_state(self, y_true, y_pred, sample_weight=None):
# for tf v1, use 'return metrics_...'. for tf v2, use 'metrics_...' (for inherited keras/tf.keras Metric class)
metrics_utils_tf_keras.update_confusion_matrix_variables(
{
metrics_utils_tf_keras.ConfusionMatrix.TRUE_POSITIVES: self.true_positives,
metrics_utils_tf_keras.ConfusionMatrix.FALSE_POSITIVES: self.false_positives,
metrics_utils_tf_keras.ConfusionMatrix.FALSE_NEGATIVES: self.false_negatives
},
y_true,
y_pred,
thresholds=self.thresholds,
top_k=self.top_k,
class_id=self.class_id,
sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
# https://github.com/tensorflow/tensorflow/issues/30711
# Remove return statement in case tensorflow.keras
update_confusion_matrix_variables(
{
ConfusionMatrix.TRUE_POSITIVES: self.true_positives,
ConfusionMatrix.TRUE_NEGATIVES: self.true_negatives,
ConfusionMatrix.FALSE_POSITIVES: self.false_positives,
ConfusionMatrix.FALSE_NEGATIVES: self.false_negatives
},
y_true,
y_pred,
thresholds=self.thresholds,
sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
"""Accumulates true negative and false positive statistics.
Args:
y_true: The ground truth values, with the same dimensions as `y_pred`.
Will be cast to `bool`.
y_pred: The predicted values. Each element must be in the range `[0, 1]`.
sample_weight: Optional weighting of each example. Defaults to 1. Can be a
`Tensor` whose rank is either 0, or the same rank as `y_true`, and must
be broadcastable to `y_true`.
Returns:
Update op.
"""
return metrics_utils.update_confusion_matrix_variables(
{
metrics_utils.ConfusionMatrix.TRUE_NEGATIVES:
self.true_negatives,
metrics_utils.ConfusionMatrix.FALSE_POSITIVES:
self.false_positives
},
y_true,
y_pred,
thresholds=self.thresholds,
top_k=self.top_k,
class_id=self.class_id,
sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
"""Accumulates true positive, false positive and false negative statistics.
Args:
y_true: The ground truth values.
y_pred: The predicted values.
sample_weight: Optional weighting of each example. Defaults to 1. Can be a
`Tensor` whose rank is either 0, or the same rank as `y_true`, and must
be broadcastable to `y_true`.
Returns:
Update op.
"""
metrics_utils.update_confusion_matrix_variables({metrics_utils.ConfusionMatrix.TRUE_POSITIVES: self.tp,
metrics_utils.ConfusionMatrix.FALSE_POSITIVES: self.fp,
metrics_utils.ConfusionMatrix.FALSE_NEGATIVES: self.fn
}, y_true, y_pred, self.thresholds,
sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
"""Accumulates the metric statistics.
Args:
y_true: The ground truth values.
y_pred: The predicted values.
sample_weight: Optional weighting of each example. Defaults to 1. Can be a
`Tensor` whose rank is either 0, or the same rank as `y_true`, and must
be broadcastable to `y_true`.
Returns:
Update op.
"""
# return
metrics_utils.update_confusion_matrix_variables(
{self._confusion_matrix_cond: self.accumulator},
y_true,
y_pred,
thresholds=self.thresholds,
class_id=self.class_id, # add class_id option here
multi_label=self.multi_label, # add multi_label option here
sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
return metrics_utils.update_confusion_matrix_variables(
{
metrics_utils.ConfusionMatrix.TRUE_POSITIVES: self.true_positives,
metrics_utils.ConfusionMatrix.FALSE_POSITIVES: self.false_positives,
metrics_utils.ConfusionMatrix.FALSE_NEGATIVES: self.false_negatives,
},
y_true=y_true,
y_pred=y_pred,
thresholds=self.thresholds,
top_k=self.top_k,
class_id=self.class_id,
sample_weight=sample_weight,
)
def update_state(self, y_true, y_pred, sample_weight=None):
if self._from_logits:
y_pred = tf.keras.activations.softmax(y_pred)
y_pred = y_pred[:, self._label]
y_true = tf.equal(self._label, tf.cast(y_true, dtype=tf.int32))
y_true = tf.reshape(y_true, shape=(-1,))
update_op = metrics_utils.update_confusion_matrix_variables({
metrics_utils.ConfusionMatrix.TRUE_POSITIVES: self.true_positives,
metrics_utils.ConfusionMatrix.TRUE_NEGATIVES: self.true_negatives,
metrics_utils.ConfusionMatrix.FALSE_POSITIVES: self.false_positives,
metrics_utils.ConfusionMatrix.FALSE_NEGATIVES: self.false_negatives,
}, y_true, y_pred, self._thresholds, sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
'''Accumulates true positive and false positive statistics.'''
if self.argmax:
y_pred = K.argmax(y_pred)
return metrics_utils.update_confusion_matrix_variables(
{
metrics_utils.ConfusionMatrix.TRUE_POSITIVES: self.true_positives,
metrics_utils.ConfusionMatrix.FALSE_POSITIVES: self.false_positives,
metrics_utils.ConfusionMatrix.FALSE_NEGATIVES: self.false_negatives
},
y_true,
y_pred,
thresholds=self.thresholds,
top_k=self.top_k,
class_id=self.class_id,
sample_weight=sample_weight)
def update_state(self, y_true, y_pred, sample_weight=None):
"""
Accumulates the given confusion matrix condition statistics.
Arguments:
y_true: The ground truth values, <tf.Tensor>.
y_pred: The predicted values, <tf.Tensor>.
sample_weight: Optional weighting of each example. Defaults to 1. Can be a
tensor whose rank is either 0, or the same rank as y_true, <tf.Tensor>.
Returns:
Update op.
"""
return metrics_utils.update_confusion_matrix_variables(
{self._confusion_matrix_cond: self.accumulator},
y_true,
y_pred,
thresholds=self.thresholds,
top_k=self.top_k,
class_id=self.class_id,
sample_weight=sample_weight,
multi_label=False,
label_weights=None)
