Python tensorflow.python.ops.math_ops.squared_difference() Examples

def setUp(self):
    super(FloatBinaryOpsTest, self).setUp()

    self.ops = [
        ('igamma', None, math_ops.igamma, core.igamma),
        ('igammac', None, math_ops.igammac, core.igammac),
        ('zeta', None, math_ops.zeta, core.zeta),
        ('polygamma', None, math_ops.polygamma, core.polygamma),
        ('maximum', None, math_ops.maximum, core.maximum),
        ('minimum', None, math_ops.minimum, core.minimum),
        ('squared_difference', None, math_ops.squared_difference,
         core.squared_difference),
    ]
    total_size = np.prod([v.size for v in self.original_lt.axes.values()])
    test_lt = core.LabeledTensor(
        math_ops.cast(self.original_lt, dtypes.float32) / total_size,
        self.original_lt.axes)
    self.test_lt_1 = test_lt
    self.test_lt_2 = 1.0 - test_lt
    self.test_lt_1_broadcast = self.test_lt_1.tensor
    self.test_lt_2_broadcast = self.test_lt_2.tensor
    self.broadcast_axes = self.test_lt_1.axes 

def setUp(self):
    super(FloatBinaryOpsTest, self).setUp()

    self.ops = [
        ('igamma', None, math_ops.igamma, core.igamma),
        ('igammac', None, math_ops.igammac, core.igammac),
        ('zeta', None, math_ops.zeta, core.zeta),
        ('polygamma', None, math_ops.polygamma, core.polygamma),
        ('maximum', None, math_ops.maximum, core.maximum),
        ('minimum', None, math_ops.minimum, core.minimum),
        ('squared_difference', None, math_ops.squared_difference,
         core.squared_difference),
    ]
    total_size = np.prod([v.size for v in self.original_lt.axes.values()])
    test_lt = core.LabeledTensor(
        math_ops.cast(self.original_lt, dtypes.float32) / total_size,
        self.original_lt.axes)
    self.test_lt_1 = test_lt
    self.test_lt_2 = 1.0 - test_lt
    self.test_lt_1_broadcast = self.test_lt_1.tensor
    self.test_lt_2_broadcast = self.test_lt_2.tensor
    self.broadcast_axes = self.test_lt_1.axes 

def mean_squared_error(predictions, labels=None, weights=1.0, scope=None):
  """Adds a Sum-of-Squares loss to the training procedure.

  `weights` acts as a coefficient for the loss. If a scalar is provided, then
  the loss is simply scaled by the given value. If `weights` is a tensor of size
  [batch_size], then the total loss for each sample of the batch is rescaled
  by the corresponding element in the `weights` vector. If the shape of
  `weights` matches the shape of `predictions`, then the loss of each
  measurable element of `predictions` is scaled by the corresponding value of
  `weights`.

  Args:
    predictions: The predicted outputs.
    labels: The ground truth output tensor, same dimensions as 'predictions'.
    weights: Coefficients for the loss a scalar, a tensor of shape
      [batch_size] or a tensor whose shape matches `predictions`.
    scope: The scope for the operations performed in computing the loss.

  Returns:
    A scalar `Tensor` representing the loss value.

  Raises:
    ValueError: If the shape of `predictions` doesn't match that of `labels` or
      if the shape of `weights` is invalid.
  """
  with ops.name_scope(scope, "mean_squared_error",
                      [predictions, labels, weights]) as scope:
    predictions.get_shape().assert_is_compatible_with(labels.get_shape())
    predictions = math_ops.cast(predictions, dtypes.float32)
    labels = math_ops.cast(labels, dtypes.float32)
    losses = math_ops.squared_difference(predictions, labels)
    return compute_weighted_loss(losses, weights, scope=scope) 

def contrastive_loss(labels, embeddings_anchor, embeddings_positive,
                     margin=1.0):
  """Computes the contrastive loss.

  This loss encourages the embedding to be close to each other for
    the samples of the same label and the embedding to be far apart at least
    by the margin constant for the samples of different labels.
  See: http://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf

  Args:
    labels: 1-D tf.int32 `Tensor` with shape [batch_size] of
      binary labels indicating positive vs negative pair.
    embeddings_anchor: 2-D float `Tensor` of embedding vectors for the anchor
      images. Embeddings should be l2 normalized.
    embeddings_positive: 2-D float `Tensor` of embedding vectors for the
      positive images. Embeddings should be l2 normalized.
    margin: margin term in the loss definition.

  Returns:
    contrastive_loss: tf.float32 scalar.
  """
  # Get per pair distances
  distances = math_ops.sqrt(
      math_ops.reduce_sum(
          math_ops.squared_difference(embeddings_anchor, embeddings_positive),
          1))

  # Add contrastive loss for the siamese network.
  #   label here is {0,1} for neg, pos.
  return math_ops.reduce_mean(
      math_ops.cast(labels, distances.dtype) * math_ops.square(distances) +
      (1. - math_ops.cast(labels, distances.dtype)) *
      math_ops.square(math_ops.maximum(margin - distances, 0.)),
      name='contrastive_loss') 

def testSquaredDifference(self):
    for dtype in [np.int32, np.float16]:
      x = np.array([[1, 2, 3], [4, 5, 6]], dtype=dtype)
      y = np.array([-3, -2, -1], dtype=dtype)
      z = (x - y)*(x - y)
      with self.test_session(use_gpu=True):
        z_tf = math_ops.squared_difference(x, y).eval()
        self.assertAllClose(z, z_tf) 

def _mean_squared_error(self, targets, outputs, mask):
        loss = math_ops.squared_difference(targets, outputs)
        # TODO: Make the below safe to div by zero
        mse = tf.reduce_sum( loss ) / tf.reduce_sum( mask )
        return mse 

def _mean_squared_error(targets, outputs, mask):
        loss = math_ops.squared_difference(targets, outputs)
        # TODO: Make the below safe to div by zero
        mse = tf.reduce_sum(loss) / tf.reduce_sum(mask)
        return mse 

def _test_squared_difference(data):
    """ One iteration of squared difference """
    return _test_elemwise(math_ops.squared_difference, data)

#######################################################################
# Floor_divide
# ------------ 

def sufficient_statistics(x, axes, shift=None, keep_dims=False, name=None):
  """Calculate the sufficient statistics for the mean and variance of `x`.

  These sufficient statistics are computed using the one pass algorithm on
  an input that's optionally shifted. See:
  https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Computing_shifted_data

  Args:
    x: A `Tensor`.
    axes: Array of ints. Axes along which to compute mean and variance.
    shift: A `Tensor` containing the value by which to shift the data for
      numerical stability, or `None` if no shift is to be performed. A shift
      close to the true mean provides the most numerically stable results.
    keep_dims: produce statistics with the same dimensionality as the input.
    name: Name used to scope the operations that compute the sufficient stats.

  Returns:
    Four `Tensor` objects of the same type as `x`:

    * the count (number of elements to average over).
    * the (possibly shifted) sum of the elements in the array.
    * the (possibly shifted) sum of squares of the elements in the array.
    * the shift by which the mean must be corrected or None if `shift` is None.
  """
  axes = list(set(axes))
  with ops.name_scope(name, "sufficient_statistics", [x, shift]):
    x = ops.convert_to_tensor(x, name="x")
    x_shape = x.get_shape()
    if all(x_shape[d].value is not None for d in axes):
      counts = 1
      for d in axes:
        counts *= x_shape[d].value
      counts = constant_op.constant(counts, dtype=x.dtype)
    else:  # shape needs to be inferred at runtime.
      x_dims = array_ops.gather(
          math_ops.cast(array_ops.shape(x), x.dtype), axes)
      counts = math_ops.reduce_prod(x_dims, name="count")
    if shift is not None:
      shift = ops.convert_to_tensor(shift, name="shift")
      m_ss = math_ops.subtract(x, shift)
      v_ss = math_ops.squared_difference(x, shift)
    else:  # no shift.
      m_ss = x
      v_ss = math_ops.square(x)
    m_ss = math_ops.reduce_sum(m_ss, axes, keep_dims=keep_dims, name="mean_ss")
    v_ss = math_ops.reduce_sum(v_ss, axes, keep_dims=keep_dims, name="var_ss")
  return counts, m_ss, v_ss, shift 

def moments(x, axes, shift=None, name=None, keep_dims=False):
  """Calculate the mean and variance of `x`.

  The mean and variance are calculated by aggregating the contents of `x`
  across `axes`.  If `x` is 1-D and `axes = [0]` this is just the mean
  and variance of a vector.

  Note: for numerical stability, when shift=None, the true mean
  would be computed and used as shift.

  When using these moments for batch normalization (see
  `tf.nn.batch_normalization`):

   * for so-called "global normalization", used with convolutional filters with
     shape `[batch, height, width, depth]`, pass `axes=[0, 1, 2]`.
   * for simple batch normalization pass `axes=[0]` (batch only).

  Args:
    x: A `Tensor`.
    axes: Array of ints.  Axes along which to compute mean and
      variance.
    shift: A `Tensor` containing the value by which to shift the data for
      numerical stability, or `None` in which case the true mean of the data is
      used as shift. A shift close to the true mean provides the most
      numerically stable results.
    name: Name used to scope the operations that compute the moments.
    keep_dims: produce moments with the same dimensionality as the input.

  Returns:
    Two `Tensor` objects: `mean` and `variance`.
  """
  with ops.name_scope(name, "moments", [x, axes, shift]):
    # The dynamic range of fp16 is too limited to support the collection of
    # sufficient statistics. As a workaround we simply perform the operations
    # on 32-bit floats before converting the mean and variance back to fp16
    y = math_ops.cast(x, dtypes.float32) if x.dtype == dtypes.float16 else x
    if shift is None:
      # Compute true mean while keeping the dims for proper broadcasting.
      shift = array_ops.stop_gradient(
          math_ops.reduce_mean(y, axes, keep_dims=True))
    else:
      shift = math_ops.cast(shift, y.dtype)
    shifted_mean = math_ops.reduce_mean(
        math_ops.subtract(y, shift), axes, keep_dims=True, name="shifted_mean")
    variance = math_ops.subtract(
        math_ops.reduce_mean(
            math_ops.squared_difference(y, shift), axes, keep_dims=True),
        math_ops.square(shifted_mean),
        name="variance")
    mean = math_ops.add(shifted_mean, shift, name="mean")
    if not keep_dims:
      mean = array_ops.squeeze(mean, axes)
      variance = array_ops.squeeze(variance, axes)
    if x.dtype == dtypes.float16:
      return (math_ops.cast(mean, dtypes.float16), math_ops.cast(
          variance, dtypes.float16))
    else:
      return (mean, variance) 

def sufficient_statistics(x, axes, shift=None, keep_dims=False, name=None):
  """Calculate the sufficient statistics for the mean and variance of `x`.

  These sufficient statistics are computed using the one pass algorithm on
  an input that's optionally shifted. See:
  https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Computing_shifted_data

  Args:
    x: A `Tensor`.
    axes: Array of ints. Axes along which to compute mean and variance.
    shift: A `Tensor` containing the value by which to shift the data for
      numerical stability, or `None` if no shift is to be performed. A shift
      close to the true mean provides the most numerically stable results.
    keep_dims: produce statistics with the same dimensionality as the input.
    name: Name used to scope the operations that compute the sufficient stats.

  Returns:
    Four `Tensor` objects of the same type as `x`:

    * the count (number of elements to average over).
    * the (possibly shifted) sum of the elements in the array.
    * the (possibly shifted) sum of squares of the elements in the array.
    * the shift by which the mean must be corrected or None if `shift` is None.
  """
  axes = list(set(axes))
  with ops.name_scope(name, "sufficient_statistics", [x, shift]):
    x = ops.convert_to_tensor(x, name="x")
    x_shape = x.get_shape()
    if all(x_shape[d].value is not None for d in axes):
      counts = 1
      for d in axes:
        counts *= x_shape[d].value
      counts = constant_op.constant(counts, dtype=x.dtype)
    else:  # shape needs to be inferred at runtime.
      x_dims = array_ops.gather(
          math_ops.cast(array_ops.shape(x), x.dtype), axes)
      counts = math_ops.reduce_prod(x_dims, name="count")
    if shift is not None:
      shift = ops.convert_to_tensor(shift, name="shift")
      m_ss = math_ops.subtract(x, shift)
      v_ss = math_ops.squared_difference(x, shift)
    else:  # no shift.
      m_ss = x
      v_ss = math_ops.square(x)
    m_ss = math_ops.reduce_sum(m_ss, axes, keep_dims=keep_dims, name="mean_ss")
    v_ss = math_ops.reduce_sum(v_ss, axes, keep_dims=keep_dims, name="var_ss")
  return counts, m_ss, v_ss, shift 

def sufficient_statistics(x, axes, shift=None, keep_dims=False, name=None):
  """Calculate the sufficient statistics for the mean and variance of `x`.

  These sufficient statistics are computed using the one pass algorithm on
  an input that's optionally shifted. See:
  https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Computing_shifted_data

  Args:
    x: A `Tensor`.
    axes: Array of ints. Axes along which to compute mean and variance.
    shift: A `Tensor` containing the value by which to shift the data for
      numerical stability, or `None` if no shift is to be performed. A shift
      close to the true mean provides the most numerically stable results.
    keep_dims: produce statistics with the same dimensionality as the input.
    name: Name used to scope the operations that compute the sufficient stats.

  Returns:
    Four `Tensor` objects of the same type as `x`:

    * the count (number of elements to average over).
    * the (possibly shifted) sum of the elements in the array.
    * the (possibly shifted) sum of squares of the elements in the array.
    * the shift by which the mean must be corrected or None if `shift` is None.
  """
  axes = list(set(axes))
  with ops.name_scope(name, "sufficient_statistics", [x, shift]):
    x = ops.convert_to_tensor(x, name="x")
    x_shape = x.get_shape()
    if x_shape.is_fully_defined():
      counts = 1
      for d in axes:
        counts *= x_shape[d].value
      counts = constant_op.constant(counts, dtype=x.dtype)
    else:  # shape needs to be inferred at runtime.
      x_dims = array_ops.gather(array_ops.shape(x), axes)
      counts = math_ops.cast(
          math_ops.reduce_prod(x_dims), x.dtype, name="count")
    if shift is not None:
      shift = ops.convert_to_tensor(shift, name="shift")
      m_ss = math_ops.sub(x, shift)
      v_ss = math_ops.squared_difference(x, shift)
    else:  # no shift.
      m_ss = x
      v_ss = math_ops.square(x)
    m_ss = math_ops.reduce_sum(m_ss, axes, keep_dims=keep_dims, name="mean_ss")
    v_ss = math_ops.reduce_sum(v_ss, axes, keep_dims=keep_dims, name="var_ss")
  return counts, m_ss, v_ss, shift 

def mean_squared_error(
    labels, predictions, weights=1.0, scope=None,
    loss_collection=ops.GraphKeys.LOSSES,
    reduction=Reduction.SUM_BY_NONZERO_WEIGHTS):
  """Adds a Sum-of-Squares loss to the training procedure.

  `weights` acts as a coefficient for the loss. If a scalar is provided, then
  the loss is simply scaled by the given value. If `weights` is a tensor of size
  [batch_size], then the total loss for each sample of the batch is rescaled
  by the corresponding element in the `weights` vector. If the shape of
  `weights` matches the shape of `predictions`, then the loss of each
  measurable element of `predictions` is scaled by the corresponding value of
  `weights`.

  Args:
    labels: The ground truth output tensor, same dimensions as 'predictions'.
    predictions: The predicted outputs.
    weights: Optional `Tensor` whose rank is either 0, or the same rank as
      `labels`, and must be broadcastable to `labels` (i.e., all dimensions must
      be either `1`, or the same as the corresponding `losses` dimension).
    scope: The scope for the operations performed in computing the loss.
    loss_collection: collection to which the loss will be added.
    reduction: Type of reduction to apply to loss.

  Returns:
    Weighted loss float `Tensor`. If `reduction` is `NONE`, this has the same
    shape as `labels`; otherwise, it is scalar.

  Raises:
    ValueError: If the shape of `predictions` doesn't match that of `labels` or
      if the shape of `weights` is invalid.  Also if `labels` or `predictions`
      is None.
  """
  if labels is None:
    raise ValueError("labels must not be None.")
  if predictions is None:
    raise ValueError("predictions must not be None.")
  with ops.name_scope(scope, "mean_squared_error",
                      (predictions, labels, weights)) as scope:
    predictions = math_ops.to_float(predictions)
    labels = math_ops.to_float(labels)
    predictions.get_shape().assert_is_compatible_with(labels.get_shape())
    losses = math_ops.squared_difference(predictions, labels)
    return compute_weighted_loss(
        losses, weights, scope, loss_collection, reduction=reduction) 

def sufficient_statistics(x, axes, shift=None, keep_dims=False, name=None):
  """Calculate the sufficient statistics for the mean and variance of `x`.

  These sufficient statistics are computed using the one pass algorithm on
  an input that's optionally shifted. See:
  https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Computing_shifted_data

  Args:
    x: A `Tensor`.
    axes: Array of ints. Axes along which to compute mean and variance.
    shift: A `Tensor` containing the value by which to shift the data for
      numerical stability, or `None` if no shift is to be performed. A shift
      close to the true mean provides the most numerically stable results.
    keep_dims: produce statistics with the same dimensionality as the input.
    name: Name used to scope the operations that compute the sufficient stats.

  Returns:
    Four `Tensor` objects of the same type as `x`:

    * the count (number of elements to average over).
    * the (possibly shifted) sum of the elements in the array.
    * the (possibly shifted) sum of squares of the elements in the array.
    * the shift by which the mean must be corrected or None if `shift` is None.
  """
  axes = list(set(axes))
  with ops.name_scope(name, "sufficient_statistics", [x, shift]):
    x = ops.convert_to_tensor(x, name="x")
    x_shape = x.get_shape()
    if all(x_shape[d].value is not None for d in axes):
      counts = 1
      for d in axes:
        counts *= x_shape[d].value
      counts = constant_op.constant(counts, dtype=x.dtype)
    else:  # shape needs to be inferred at runtime.
      x_dims = array_ops.gather(
          math_ops.cast(array_ops.shape(x), x.dtype), axes)
      counts = math_ops.reduce_prod(x_dims, name="count")
    if shift is not None:
      shift = ops.convert_to_tensor(shift, name="shift")
      m_ss = math_ops.subtract(x, shift)
      v_ss = math_ops.squared_difference(x, shift)
    else:  # no shift.
      m_ss = x
      v_ss = math_ops.square(x)
    m_ss = math_ops.reduce_sum(m_ss, axes, keep_dims=keep_dims, name="mean_ss")
    v_ss = math_ops.reduce_sum(v_ss, axes, keep_dims=keep_dims, name="var_ss")
  return counts, m_ss, v_ss, shift 

def moments(x, axes,
            shift=None,  # pylint: disable=unused-argument
            name=None, keep_dims=False):
  """Calculate the mean and variance of `x`.

  The mean and variance are calculated by aggregating the contents of `x`
  across `axes`.  If `x` is 1-D and `axes = [0]` this is just the mean
  and variance of a vector.

  Note: shift is currently not used, the true mean is computed and used.

  When using these moments for batch normalization (see
  `tf.nn.batch_normalization`):

   * for so-called "global normalization", used with convolutional filters with
     shape `[batch, height, width, depth]`, pass `axes=[0, 1, 2]`.
   * for simple batch normalization pass `axes=[0]` (batch only).

  Args:
    x: A `Tensor`.
    axes: Array of ints.  Axes along which to compute mean and
      variance.
    shift: Not used in the current implementation
    name: Name used to scope the operations that compute the moments.
    keep_dims: produce moments with the same dimensionality as the input.

  Returns:
    Two `Tensor` objects: `mean` and `variance`.
  """
  with ops.name_scope(name, "moments", [x, axes]):
    # The dynamic range of fp16 is too limited to support the collection of
    # sufficient statistics. As a workaround we simply perform the operations
    # on 32-bit floats before converting the mean and variance back to fp16
    y = math_ops.cast(x, dtypes.float32) if x.dtype == dtypes.float16 else x
    # Compute true mean while keeping the dims for proper broadcasting.
    mean = math_ops.reduce_mean(y, axes, keep_dims=True, name="mean")
    # sample variance, not unbiased variance
    variance = math_ops.reduce_mean(
        math_ops.squared_difference(y, array_ops.stop_gradient(mean)),
        axes,
        keep_dims=True,
        name="variance")
    if not keep_dims:
      mean = array_ops.squeeze(mean, axes)
      variance = array_ops.squeeze(variance, axes)
    if x.dtype == dtypes.float16:
      return (math_ops.cast(mean, dtypes.float16), math_ops.cast(
          variance, dtypes.float16))
    else:
      return (mean, variance) 

def sufficient_statistics(x, axes, shift=None, keep_dims=False, name=None):
  """Calculate the sufficient statistics for the mean and variance of `x`.

  These sufficient statistics are computed using the one pass algorithm on
  an input that's optionally shifted. See:
  https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Computing_shifted_data

  Args:
    x: A `Tensor`.
    axes: Array of ints. Axes along which to compute mean and variance.
    shift: A `Tensor` containing the value by which to shift the data for
      numerical stability, or `None` if no shift is to be performed. A shift
      close to the true mean provides the most numerically stable results.
    keep_dims: produce statistics with the same dimensionality as the input.
    name: Name used to scope the operations that compute the sufficient stats.

  Returns:
    Four `Tensor` objects of the same type as `x`:

    * the count (number of elements to average over).
    * the (possibly shifted) sum of the elements in the array.
    * the (possibly shifted) sum of squares of the elements in the array.
    * the shift by which the mean must be corrected or None if `shift` is None.
  """
  axes = list(set(axes))
  with ops.name_scope(name, "sufficient_statistics", [x, shift]):
    x = ops.convert_to_tensor(x, name="x")
    x_shape = x.get_shape()
    if all(x_shape[d].value is not None for d in axes):
      counts = 1
      for d in axes:
        counts *= x_shape[d].value
      counts = constant_op.constant(counts, dtype=x.dtype)
    else:  # shape needs to be inferred at runtime.
      x_dims = array_ops.gather(
          math_ops.cast(array_ops.shape(x), x.dtype), axes)
      counts = math_ops.reduce_prod(x_dims, name="count")
    if shift is not None:
      shift = ops.convert_to_tensor(shift, name="shift")
      m_ss = math_ops.subtract(x, shift)
      v_ss = math_ops.squared_difference(x, shift)
    else:  # no shift.
      m_ss = x
      v_ss = math_ops.square(x)
    m_ss = math_ops.reduce_sum(m_ss, axes, keep_dims=keep_dims, name="mean_ss")
    v_ss = math_ops.reduce_sum(v_ss, axes, keep_dims=keep_dims, name="var_ss")
  return counts, m_ss, v_ss, shift