npdl.objectives
¶
Provides some minimal help with building loss expressions for training or validating a neural network.
These functions build element- or item-wise loss expressions from network predictions and targets.
Examples¶
Assuming you have a simple neural network for 3-way classification:
>>> import npdl
>>> model = npdl.model.Model()
>>> model.add(npdl.layers.Dense(n_out=100, n_in=50))
>>> model.add(npdl.layers.Dense(n_out=3, activation=npdl.activations.Softmax()))
>>> model.compile(loss=npdl.objectives.SCCE(), optimizer=npdl.optimizers.SGD(lr=0.005))
Objectives¶
MeanSquaredError |
Computes the element-wise squared difference between targets and outputs . |
MSE |
alias of MeanSquaredError |
HellingerDistance |
Computes the multi-class hinge loss between predictions and targets. |
HeD |
alias of HellingerDistance |
BinaryCrossEntropy |
Computes the binary cross-entropy between predictions and targets. |
BCE |
alias of BinaryCrossEntropy |
SoftmaxCategoricalCrossEntropy |
Computes the categorical cross-entropy between predictions and targets. |
SCCE |
alias of SoftmaxCategoricalCrossEntropy |
Detailed Description¶
-
class
npdl.objectives.
Objective
[source]¶ An objective function (or loss function, or optimization score function) is one of the two parameters required to compile a model.
-
class
npdl.objectives.
MeanSquaredError
[source]¶ Computes the element-wise squared difference between
targets
andoutputs
.In statistics, the mean squared error (MSE) or mean squared deviation (MSD) of an estimator (of a procedure for estimating an unobserved quantity) measures the average of the squares of the errors or deviations—that is, the difference between the estimator and what is estimated. MSE is a risk function, corresponding to the expected value of the squared error loss or quadratic loss. The difference occurs because of randomness or because the estimator doesn’t account for information that could produce a more accurate estimate. [R4747]
The MSE is a measure of the quality of an estimator—it is always non-negative, and values closer to zero are better.
The MSE is the second moment (about the origin) of the error, and thus incorporates both the variance of the estimator and its bias. For an unbiased estimator, the MSE is the variance of the estimator. Like the variance, MSE has the same units of measurement as the square of the quantity being estimated. In an analogy to standard deviation, taking the square root of MSE yields the root-mean-square error or root-mean-square deviation (RMSE or RMSD), which has the same units as the quantity being estimated; for an unbiased estimator, the RMSE is the square root of the variance, known as the standard deviation.
Notes
This is the loss function of choice for many regression problems or auto-encoders with linear output units.
References
[R4747] (1, 2) Lehmann, E. L.; Casella, George (1998). Theory of Point Estimation (2nd ed.). New York: Springer. ISBN 0-387-98502-6. MR 1639875.
-
npdl.objectives.
MSE
[source]¶ alias of
MeanSquaredError
-
class
npdl.objectives.
HellingerDistance
[source]¶ Computes the multi-class hinge loss between predictions and targets.
In probability and statistics, the Hellinger distance (closely related to, although different from, the Bhattacharyya distance) is used to quantify the similarity between two probability distributions. It is a type of f-divergence. The Hellinger distance is defined in terms of the Hellinger integral, which was introduced by Ernst Hellinger in 1909.[R4950]_ [R5050]
Notes
This is an alternative to the categorical cross-entropy loss for multi-class classification problems
References
[R4950] Nikulin, M.S. (2001), “Hellinger distance”, in Hazewinkel, Michiel, Encyclopedia of Mathematics, Springer, ISBN 978-1-55608-010-4 [R5050] (1, 2) Jump up ^ Hellinger, Ernst (1909), “Neue Begründung der Theorie quadratischer Formen von unendlichvielen Veränderlichen”, Journal für die reine und angewandte Mathematik (in German), 136: 210–271, doi:10.1515/crll.1909.136.210, JFM 40.0393.01 -
forward
(outputs, targets)[source]¶ HellingerDistance forward propagation.
Parameters: outputs : numpy 2D array
outputs in (0, 1), such as softmax output of a neural network, with data points in rows and class probabilities in columns.
targets : numpy 2D array
Either a vector of int giving the correct class index per data point or a 2D tensor of one-hot encoding of the correct class in the same layout as predictions (non-binary targets in [0, 1] do not work!)
Returns: numpy 1D array
An expression for the Hellinger Distance
-
-
npdl.objectives.
HeD
[source]¶ alias of
HellingerDistance
-
class
npdl.objectives.
BinaryCrossEntropy
(epsilon=1e-11)[source]¶ Computes the binary cross-entropy between predictions and targets.
Returns: numpy array
An expression for the element-wise binary cross-entropy.
Notes
This is the loss function of choice for binary classification problems and sigmoid output units.
-
npdl.objectives.
BCE
[source]¶ alias of
BinaryCrossEntropy
-
class
npdl.objectives.
SoftmaxCategoricalCrossEntropy
(epsilon=1e-11)[source]¶ Computes the categorical cross-entropy between predictions and targets.
Notes
This is the loss function of choice for multi-class classification problems and softmax output units. For hard targets, i.e., targets that assign all of the probability to a single class per data point, providing a vector of int for the targets is usually slightly more efficient than providing a matrix with a single 1.0 per row.
-
backward
(outputs, targets)[source]¶ SoftmaxCategoricalCrossEntropy backward propagation.
\[dE = p - t\]Parameters: outputs : numpy 2D array
Predictions in (0, 1), such as softmax output of a neural network, with data points in rows and class probabilities in columns.
targets : numpy 2D array
Either targets in [0, 1] matching the layout of outputs, or a vector of int giving the correct class index per data point.
Returns: numpy 1D array
-
forward
(outputs, targets)[source]¶ SoftmaxCategoricalCrossEntropy forward propagation.
\[L_i = - \sum_j{t_{i,j} \log(p_{i,j})}\]Parameters: outputs : numpy.array
Predictions in (0, 1), such as softmax output of a neural network, with data points in rows and class probabilities in columns.
targets : numpy.array
Either targets in [0, 1] matching the layout of outputs, or a vector of int giving the correct class index per data point.
Returns: numpy 1D array
An expression for the item-wise categorical cross-entropy.
-
-
npdl.objectives.
SCCE
[source]¶ alias of
SoftmaxCategoricalCrossEntropy