Add a PrecisionAtRecall metric.

PiperOrigin-RevId: 274973353
Change-Id: I4a64d6127a0bf9281627f7a3de8cc3fccca882dd

tensorflow/python/keras/metrics.py

@@ -1556,7 +1556,7 @@ class SpecificityAtSensitivity(SensitivitySpecificityBase):
     Args:
       sensitivity: A scalar value in range `[0, 1]`.
       num_thresholds: (Optional) Defaults to 200. The number of thresholds to
-        use for matching the given specificity.
+        use for matching the given sensitivity.
       name: (Optional) string name of the metric instance.
       dtype: (Optional) data type of the metric result.
     """
@@ -1573,7 +1573,7 @@ class SpecificityAtSensitivity(SensitivitySpecificityBase):
         self.true_positives, self.true_positives + self.false_negatives)
 
     # Find the index of the threshold where the sensitivity is closest to the
-    # given specificity.
+    # requested value.
     min_index = math_ops.argmin(
         math_ops.abs(sensitivities - self.value), axis=0)
     min_index = math_ops.cast(min_index, dtypes.int32)
@@ -1592,6 +1592,79 @@ class SpecificityAtSensitivity(SensitivitySpecificityBase):
     return dict(list(base_config.items()) + list(config.items()))
 
 
+@keras_export('keras.metrics.PrecisionAtRecall')
+class PrecisionAtRecall(SensitivitySpecificityBase):
+  """Computes the precision at a given recall.
+
+  This metric creates four local variables, `true_positives`, `true_negatives`,
+  `false_positives` and `false_negatives` that are used to compute the
+  precision at the given recall. The threshold for the given recall
+  value is computed and used to evaluate the corresponding precision.
+
+  If `sample_weight` is `None`, weights default to 1.
+  Use `sample_weight` of 0 to mask values.
+
+  Usage:
+
+  ```python
+  m = tf.keras.metrics.PrecisionAtRecall(0.8, num_thresholds=1)
+  m.update_state([0, 0, 1, 1], [0, 0.5, 0.3, 0.9])
+  print('Final result: ', m.result().numpy())  # Final result: 1.0
+  ```
+
+  Usage with tf.keras API:
+
+  ```python
+  model = tf.keras.Model(inputs, outputs)
+  model.compile(
+      'sgd',
+      loss='mse',
+      metrics=[tf.keras.metrics.PrecisionAtRecall()])
+  ```
+  """
+
+  def __init__(self, recall, num_thresholds=200, name=None, dtype=None):
+    """Creates a `PrecisionAtRecall` instance.
+
+    Args:
+      recall: A scalar value in range `[0, 1]`.
+      num_thresholds: (Optional) Defaults to 200. The number of thresholds to
+        use for matching the given recall.
+      name: (Optional) string name of the metric instance.
+      dtype: (Optional) data type of the metric result.
+    """
+    if recall < 0 or recall > 1:
+      raise ValueError('`recall` must be in the range [0, 1].')
+    self.recall = recall
+    self.num_thresholds = num_thresholds
+    super(PrecisionAtRecall, self).__init__(
+        value=recall,
+        num_thresholds=num_thresholds,
+        name=name,
+        dtype=dtype)
+
+  def result(self):
+    # Calculate recall at all the thresholds.
+    recalls = math_ops.div_no_nan(
+        self.true_positives, self.true_positives + self.false_negatives)
+
+    # Find the index of the threshold where the recall is closest to the
+    # requested value.
+    min_index = math_ops.argmin(
+        math_ops.abs(recalls - self.value), axis=0)
+    min_index = math_ops.cast(min_index, dtypes.int32)
+
+    # Compute precision at that index.
+    return math_ops.div_no_nan(
+        self.true_positives[min_index],
+        self.true_positives[min_index] + self.false_positives[min_index])
+
+  def get_config(self):
+    config = {'num_thresholds': self.num_thresholds, 'recall': self.recall}
+    base_config = super(PrecisionAtRecall, self).get_config()
+    return dict(list(base_config.items()) + list(config.items()))
+
+
 @keras_export('keras.metrics.AUC')
 class AUC(Metric):
   """Computes the approximate AUC (Area under the curve) via a Riemann sum.

tensorflow/python/keras/metrics_confusion_matrix_test.py

@@ -919,6 +919,103 @@ class SpecificityAtSensitivityTest(test.TestCase, parameterized.TestCase):
       metrics.SpecificityAtSensitivity(0.4, num_thresholds=-1)
 
 
+@test_util.run_all_in_graph_and_eager_modes
+class PrecisionAtRecallTest(test.TestCase, parameterized.TestCase):
+
+  def test_config(self):
+    s_obj = metrics.PrecisionAtRecall(
+        0.4, num_thresholds=100, name='precision_at_recall_1')
+    self.assertEqual(s_obj.name, 'precision_at_recall_1')
+    self.assertLen(s_obj.variables, 4)
+    self.assertEqual(s_obj.recall, 0.4)
+    self.assertEqual(s_obj.num_thresholds, 100)
+
+    # Check save and restore config
+    s_obj2 = metrics.PrecisionAtRecall.from_config(s_obj.get_config())
+    self.assertEqual(s_obj2.name, 'precision_at_recall_1')
+    self.assertLen(s_obj2.variables, 4)
+    self.assertEqual(s_obj2.recall, 0.4)
+    self.assertEqual(s_obj2.num_thresholds, 100)
+
+  def test_value_is_idempotent(self):
+    s_obj = metrics.PrecisionAtRecall(0.7)
+    y_pred = random_ops.random_uniform((10, 3),
+                                       maxval=1,
+                                       dtype=dtypes.float32,
+                                       seed=1)
+    y_true = random_ops.random_uniform((10, 3),
+                                       maxval=2,
+                                       dtype=dtypes.int64,
+                                       seed=1)
+    update_op = s_obj.update_state(y_true, y_pred)
+    self.evaluate(variables.variables_initializer(s_obj.variables))
+
+    # Run several updates.
+    for _ in range(10):
+      self.evaluate(update_op)
+
+    # Then verify idempotency.
+    initial_specificity = self.evaluate(s_obj.result())
+    for _ in range(10):
+      self.assertAlmostEqual(initial_specificity, self.evaluate(s_obj.result()),
+                             1e-3)
+
+  def test_unweighted_all_correct(self):
+    s_obj = metrics.PrecisionAtRecall(0.7)
+    inputs = np.random.randint(0, 2, size=(100, 1))
+    y_pred = constant_op.constant(inputs, dtype=dtypes.float32)
+    y_true = constant_op.constant(inputs)
+    self.evaluate(variables.variables_initializer(s_obj.variables))
+    result = s_obj(y_true, y_pred)
+    self.assertAlmostEqual(1, self.evaluate(result))
+
+  def test_unweighted_high_recall(self):
+    s_obj = metrics.PrecisionAtRecall(0.8)
+    pred_values = [0.0, 0.1, 0.2, 0.3, 0.5, 0.4, 0.5, 0.6, 0.8, 0.9]
+    label_values = [0, 0, 0, 0, 0, 1, 1, 1, 1, 1]
+
+    # For a score between 0.4 and 0.5, we expect 0.8 precision, 0.8 recall.
+    y_pred = constant_op.constant(pred_values, dtype=dtypes.float32)
+    y_true = constant_op.constant(label_values)
+    self.evaluate(variables.variables_initializer(s_obj.variables))
+    result = s_obj(y_true, y_pred)
+    self.assertAlmostEqual(0.8, self.evaluate(result))
+
+  def test_unweighted_low_recall(self):
+    s_obj = metrics.PrecisionAtRecall(0.4)
+    pred_values = [0.0, 0.1, 0.2, 0.3, 0.4, 0.1, 0.15, 0.25, 0.26, 0.26]
+    label_values = [0, 0, 0, 0, 0, 1, 1, 1, 1, 1]
+
+    y_pred = constant_op.constant(pred_values, dtype=dtypes.float32)
+    y_true = constant_op.constant(label_values)
+    self.evaluate(variables.variables_initializer(s_obj.variables))
+    result = s_obj(y_true, y_pred)
+    self.assertAlmostEqual(0.5, self.evaluate(result))
+
+  @parameterized.parameters([dtypes.bool, dtypes.int32, dtypes.float32])
+  def test_weighted(self, label_dtype):
+    s_obj = metrics.PrecisionAtRecall(0.4)
+    pred_values = [0.0, 0.1, 0.2, 0.3, 0.4, 0.01, 0.02, 0.25, 0.26, 0.26]
+    label_values = [0, 0, 0, 0, 0, 1, 1, 1, 1, 1]
+    weight_values = [2, 2, 1, 1, 1, 1, 1, 2, 2, 2]
+
+    y_pred = constant_op.constant(pred_values, dtype=dtypes.float32)
+    y_true = math_ops.cast(label_values, dtype=label_dtype)
+    weights = constant_op.constant(weight_values)
+    self.evaluate(variables.variables_initializer(s_obj.variables))
+    result = s_obj(y_true, y_pred, sample_weight=weights)
+    self.assertAlmostEqual(2./3., self.evaluate(result))
+
+  def test_invalid_sensitivity(self):
+    with self.assertRaisesRegexp(
+        ValueError, r'`recall` must be in the range \[0, 1\].'):
+      metrics.PrecisionAtRecall(-1)
+
+  def test_invalid_num_thresholds(self):
+    with self.assertRaisesRegexp(ValueError, '`num_thresholds` must be > 0.'):
+      metrics.PrecisionAtRecall(0.4, num_thresholds=-1)
+
+
 @test_util.run_all_in_graph_and_eager_modes
 class AUCTest(test.TestCase):
 

tensorflow/tools/api/golden/v1/tensorflow.keras.metrics.-precision-at-recall.pbtxt

@@ -0,0 +1,231 @@
+path: "tensorflow.keras.metrics.PrecisionAtRecall"
+tf_class {
+  is_instance: "<class \'tensorflow.python.keras.metrics.PrecisionAtRecall\'>"
+  is_instance: "<class \'tensorflow.python.keras.metrics.SensitivitySpecificityBase\'>"
+  is_instance: "<class \'tensorflow.python.keras.metrics.Metric\'>"
+  is_instance: "<class \'tensorflow.python.keras.engine.base_layer.Layer\'>"
+  is_instance: "<class \'tensorflow.python.module.module.Module\'>"
+  is_instance: "<class \'tensorflow.python.training.tracking.tracking.AutoTrackable\'>"
+  is_instance: "<class \'tensorflow.python.training.tracking.base.Trackable\'>"
+  is_instance: "<type \'object\'>"
+  member {
+    name: "activity_regularizer"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "dtype"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "dynamic"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "inbound_nodes"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_mask"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_shape"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_spec"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "losses"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "metrics"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "name"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "name_scope"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "non_trainable_variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "non_trainable_weights"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "outbound_nodes"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output_mask"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output_shape"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "stateful"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "submodules"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable_variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable_weights"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "updates"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "weights"
+    mtype: "<type \'property\'>"
+  }
+  member_method {
+    name: "__init__"
+    argspec: "args=[\'self\', \'recall\', \'num_thresholds\', \'name\', \'dtype\'], varargs=None, keywords=None, defaults=[\'200\', \'None\', \'None\'], "
+  }
+  member_method {
+    name: "add_loss"
+    argspec: "args=[\'self\', \'losses\', \'inputs\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "add_metric"
+    argspec: "args=[\'self\', \'value\', \'aggregation\', \'name\'], varargs=None, keywords=None, defaults=[\'None\', \'None\'], "
+  }
+  member_method {
+    name: "add_update"
+    argspec: "args=[\'self\', \'updates\', \'inputs\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "add_variable"
+    argspec: "args=[\'self\'], varargs=args, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "add_weight"
+    argspec: "args=[\'self\', \'name\', \'shape\', \'aggregation\', \'synchronization\', \'initializer\', \'dtype\'], varargs=None, keywords=None, defaults=[\'()\', \'VariableAggregation.SUM\', \'VariableSynchronization.ON_READ\', \'None\', \'None\'], "
+  }
+  member_method {
+    name: "apply"
+    argspec: "args=[\'self\', \'inputs\'], varargs=args, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "build"
+    argspec: "args=[\'self\', \'input_shape\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "call"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "compute_mask"
+    argspec: "args=[\'self\', \'inputs\', \'mask\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "compute_output_shape"
+    argspec: "args=[\'self\', \'input_shape\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "compute_output_signature"
+    argspec: "args=[\'self\', \'input_signature\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "count_params"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "from_config"
+    argspec: "args=[\'cls\', \'config\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_config"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_mask_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_shape_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_losses_for"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_mask_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_shape_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_updates_for"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_weights"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "reset_states"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "result"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "set_weights"
+    argspec: "args=[\'self\', \'weights\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "update_state"
+    argspec: "args=[\'self\', \'y_true\', \'y_pred\', \'sample_weight\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "with_name_scope"
+    argspec: "args=[\'cls\', \'method\'], varargs=None, keywords=None, defaults=None"
+  }
+}

tensorflow/tools/api/golden/v1/tensorflow.keras.metrics.pbtxt

@@ -96,6 +96,10 @@ tf_module {
     name: "Precision"
     mtype: "<type \'type\'>"
   }
+  member {
+    name: "PrecisionAtRecall"
+    mtype: "<type \'type\'>"
+  }
   member {
     name: "Recall"
     mtype: "<type \'type\'>"

tensorflow/tools/api/golden/v2/tensorflow.keras.metrics.-precision-at-recall.pbtxt

@@ -0,0 +1,231 @@
+path: "tensorflow.keras.metrics.PrecisionAtRecall"
+tf_class {
+  is_instance: "<class \'tensorflow.python.keras.metrics.PrecisionAtRecall\'>"
+  is_instance: "<class \'tensorflow.python.keras.metrics.SensitivitySpecificityBase\'>"
+  is_instance: "<class \'tensorflow.python.keras.metrics.Metric\'>"
+  is_instance: "<class \'tensorflow.python.keras.engine.base_layer.Layer\'>"
+  is_instance: "<class \'tensorflow.python.module.module.Module\'>"
+  is_instance: "<class \'tensorflow.python.training.tracking.tracking.AutoTrackable\'>"
+  is_instance: "<class \'tensorflow.python.training.tracking.base.Trackable\'>"
+  is_instance: "<type \'object\'>"
+  member {
+    name: "activity_regularizer"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "dtype"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "dynamic"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "inbound_nodes"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_mask"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_shape"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_spec"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "losses"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "metrics"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "name"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "name_scope"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "non_trainable_variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "non_trainable_weights"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "outbound_nodes"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output_mask"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output_shape"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "stateful"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "submodules"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable_variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable_weights"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "updates"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "weights"
+    mtype: "<type \'property\'>"
+  }
+  member_method {
+    name: "__init__"
+    argspec: "args=[\'self\', \'recall\', \'num_thresholds\', \'name\', \'dtype\'], varargs=None, keywords=None, defaults=[\'200\', \'None\', \'None\'], "
+  }
+  member_method {
+    name: "add_loss"
+    argspec: "args=[\'self\', \'losses\', \'inputs\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "add_metric"
+    argspec: "args=[\'self\', \'value\', \'aggregation\', \'name\'], varargs=None, keywords=None, defaults=[\'None\', \'None\'], "
+  }
+  member_method {
+    name: "add_update"
+    argspec: "args=[\'self\', \'updates\', \'inputs\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "add_variable"
+    argspec: "args=[\'self\'], varargs=args, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "add_weight"
+    argspec: "args=[\'self\', \'name\', \'shape\', \'aggregation\', \'synchronization\', \'initializer\', \'dtype\'], varargs=None, keywords=None, defaults=[\'()\', \'VariableAggregation.SUM\', \'VariableSynchronization.ON_READ\', \'None\', \'None\'], "
+  }
+  member_method {
+    name: "apply"
+    argspec: "args=[\'self\', \'inputs\'], varargs=args, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "build"
+    argspec: "args=[\'self\', \'input_shape\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "call"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "compute_mask"
+    argspec: "args=[\'self\', \'inputs\', \'mask\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "compute_output_shape"
+    argspec: "args=[\'self\', \'input_shape\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "compute_output_signature"
+    argspec: "args=[\'self\', \'input_signature\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "count_params"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "from_config"
+    argspec: "args=[\'cls\', \'config\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_config"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_mask_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_shape_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_losses_for"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_mask_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_shape_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_updates_for"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_weights"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "reset_states"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "result"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "set_weights"
+    argspec: "args=[\'self\', \'weights\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "update_state"
+    argspec: "args=[\'self\', \'y_true\', \'y_pred\', \'sample_weight\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "with_name_scope"
+    argspec: "args=[\'cls\', \'method\'], varargs=None, keywords=None, defaults=None"
+  }
+}

tensorflow/tools/api/golden/v2/tensorflow.keras.metrics.pbtxt

@@ -96,6 +96,10 @@ tf_module {
     name: "Precision"
     mtype: "<type \'type\'>"
   }
+  member {
+    name: "PrecisionAtRecall"
+    mtype: "<type \'type\'>"
+  }
   member {
     name: "Recall"
     mtype: "<type \'type\'>"

tensorflow/tools/api/golden/v2/tensorflow.metrics.-precision-at-recall.pbtxt

@@ -0,0 +1,231 @@
+path: "tensorflow.metrics.PrecisionAtRecall"
+tf_class {
+  is_instance: "<class \'tensorflow.python.keras.metrics.PrecisionAtRecall\'>"
+  is_instance: "<class \'tensorflow.python.keras.metrics.SensitivitySpecificityBase\'>"
+  is_instance: "<class \'tensorflow.python.keras.metrics.Metric\'>"
+  is_instance: "<class \'tensorflow.python.keras.engine.base_layer.Layer\'>"
+  is_instance: "<class \'tensorflow.python.module.module.Module\'>"
+  is_instance: "<class \'tensorflow.python.training.tracking.tracking.AutoTrackable\'>"
+  is_instance: "<class \'tensorflow.python.training.tracking.base.Trackable\'>"
+  is_instance: "<type \'object\'>"
+  member {
+    name: "activity_regularizer"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "dtype"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "dynamic"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "inbound_nodes"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_mask"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_shape"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "input_spec"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "losses"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "metrics"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "name"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "name_scope"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "non_trainable_variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "non_trainable_weights"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "outbound_nodes"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output_mask"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "output_shape"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "stateful"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "submodules"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable_variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "trainable_weights"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "updates"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "variables"
+    mtype: "<type \'property\'>"
+  }
+  member {
+    name: "weights"
+    mtype: "<type \'property\'>"
+  }
+  member_method {
+    name: "__init__"
+    argspec: "args=[\'self\', \'recall\', \'num_thresholds\', \'name\', \'dtype\'], varargs=None, keywords=None, defaults=[\'200\', \'None\', \'None\'], "
+  }
+  member_method {
+    name: "add_loss"
+    argspec: "args=[\'self\', \'losses\', \'inputs\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "add_metric"
+    argspec: "args=[\'self\', \'value\', \'aggregation\', \'name\'], varargs=None, keywords=None, defaults=[\'None\', \'None\'], "
+  }
+  member_method {
+    name: "add_update"
+    argspec: "args=[\'self\', \'updates\', \'inputs\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "add_variable"
+    argspec: "args=[\'self\'], varargs=args, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "add_weight"
+    argspec: "args=[\'self\', \'name\', \'shape\', \'aggregation\', \'synchronization\', \'initializer\', \'dtype\'], varargs=None, keywords=None, defaults=[\'()\', \'VariableAggregation.SUM\', \'VariableSynchronization.ON_READ\', \'None\', \'None\'], "
+  }
+  member_method {
+    name: "apply"
+    argspec: "args=[\'self\', \'inputs\'], varargs=args, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "build"
+    argspec: "args=[\'self\', \'input_shape\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "call"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=kwargs, defaults=None"
+  }
+  member_method {
+    name: "compute_mask"
+    argspec: "args=[\'self\', \'inputs\', \'mask\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "compute_output_shape"
+    argspec: "args=[\'self\', \'input_shape\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "compute_output_signature"
+    argspec: "args=[\'self\', \'input_signature\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "count_params"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "from_config"
+    argspec: "args=[\'cls\', \'config\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_config"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_mask_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_input_shape_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_losses_for"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_mask_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_output_shape_at"
+    argspec: "args=[\'self\', \'node_index\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_updates_for"
+    argspec: "args=[\'self\', \'inputs\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "get_weights"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "reset_states"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "result"
+    argspec: "args=[\'self\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "set_weights"
+    argspec: "args=[\'self\', \'weights\'], varargs=None, keywords=None, defaults=None"
+  }
+  member_method {
+    name: "update_state"
+    argspec: "args=[\'self\', \'y_true\', \'y_pred\', \'sample_weight\'], varargs=None, keywords=None, defaults=[\'None\'], "
+  }
+  member_method {
+    name: "with_name_scope"
+    argspec: "args=[\'cls\', \'method\'], varargs=None, keywords=None, defaults=None"
+  }
+}

tensorflow/tools/api/golden/v2/tensorflow.metrics.pbtxt

@@ -96,6 +96,10 @@ tf_module {
     name: "Precision"
     mtype: "<type \'type\'>"
   }
+  member {
+    name: "PrecisionAtRecall"
+    mtype: "<type \'type\'>"
+  }
   member {
     name: "Recall"
     mtype: "<type \'type\'>"