objectives.py

"""Loss functions."""

import tensorflow as tf
import semver


def huber_loss(y_true, y_pred, max_grad=1.):
    """Calculate the huber loss.

    See https://en.wikipedia.org/wiki/Huber_loss

    Parameters
    ----------
    y_true: np.array, tf.Tensor
      Target value.
    y_pred: np.array, tf.Tensor
      Predicted value.
    max_grad: float, optional
      Positive floating point value. Represents the maximum possible
      gradient magnitude.

    Returns
    -------
    tf.Tensor
      The huber loss.
    """
    residual=tf.abs(y_pred-y_true)
    condition=tf.less(residual,max_grad)
    small_residual=0.5*tf.square(residual)
    large_residual=max_grad*residual-0.5*tf.square(max_grad)
    return tf.select(condition,small_residual,large_residual)


def mean_huber_loss(y_true, y_pred, max_grad=1.):
    """Return mean huber loss.

    Same as huber_loss, but takes the mean over all values in the
    output tensor.

    Parameters
    ----------
    y_true: np.array, tf.Tensor
      Target value.
    y_pred: np.array, tf.Tensor
      Predicted value.
    max_grad: float, optional
      Positive floating point value. Represents the maximum possible
      gradient magnitude.

    Returns
    -------
    tf.Tensor
      The mean huber loss.
    """
    loss_tensor=huber_loss(y_true,y_pred,max_grad)
    return tf.reduce_mean(loss_tensor)