Developer Guide and Reference

  • 2021.1
  • 12/04/2020
  • Public Content
Contents

Multi-class Classifier

While some classification algorithms naturally permit the use of more than two classes, some algorithms, such as Support Vector Machines (SVM), are by nature solving a two-class problem only. These two-class (or binary) classifiers can be turned into multi-class classifiers by using different strategies, such as One-Against-Rest or One-Against-One.
oneDAL implements a Multi-Class Classifier using the One-Against-One strategy.
Multi-class classifiers, such as SVM, are based on two-class classifiers, which are integral components of the models trained with the corresponding multi-class classifier algorithms.

Details

Given
n
feature vectors LaTex Math image. of size
p
, the number of classes
K
, and a vector of class labels LaTex Math image. , where LaTex Math image. , the problem is to build a multi-class classifier using a two-class (binary) classifier, such as a two-class SVM.
Training Stage
The model is trained with the One-Against-One method that uses the binary classification described in [Hsu02] as follows: For each pair of classes LaTex Math image. , train a binary classifier, such as SVM. The total number of such binary classifiers is LaTex Math image. .
Prediction Stage
Given a new feature vector LaTex Math image. , the classifier determines the class to which the vector belongs.
oneDAL provides two methods for class label prediction:
  • Wu method. According to the algorithm 2 for computation of the class probabilities described in [Wu04]. The library returns the index of the class with the largest probability.
  • Vote-based method. If the binary classifier predicts the feature vector to be in
    i
    -th class, the number of votes for the class i is increased by one, otherwise the vote is given to the j-th class. If two classes have equal numbers of votes, the class with the smallest index is selected.

Usage of Training Alternative

To build a Multi-class Classifier model using methods of the Model Builder class of Multi-class Classifier, complete the following steps:
  • Create a Multi-class Classifier model builder using a constructor with the required number of features and classes.
  • Use the
    setTwoClassClassifierModel
    method for each pair of classes to add the pre-trained two-class classifiers to the model. In the parameters to the method specify the classes’ indices and the pointer to the pre-trained two-class classifier for this pair of classes. You need to do this for each pair of classes, because the One-Against-One strategy is used.
  • Use the
    getModel
    method to get the trained Multi-class Classifier model.
  • Use the
    getStatus
    method to check the status of the model building process. If
    DAAL_NOTHROW_EXCEPTIONS
    macros is defined, the status report contains the list of errors that describe the problems API encountered (in case of API runtime failure).
Examples
oneAPI C++
C++ (CPU)
Java*
There is no support for Java on GPU.
Batch Processing

Batch Processing

Multi-class classifier follows the general workflow described in Classification Usage Model.
Training
At the training stage, a multi-class classifier has the following parameters:
Parameter
Default Value
Description
algorithmFPType
float
The floating-point type that the algorithm uses for intermediate computations. Can be
float
or
double
.
method
defaultDense
The computation method used by the multi-class classifier. The only training method supported so far is One-Against-One.
training
Pointer to an object of the SVM training class
Pointer to the training algorithm of the two-class classifier. By default, the SVM two-class classifier is used.
nClasses
Not applicable
The number of classes. A required parameter.
Prediction
At the prediction stage, a multi-class classifier has the following parameters:
Parameter
Method
Default Value
Description
algorithmFPType
defaultDense
or
voteBased
float
The floating-point type that the algorithm uses for intermediate computations. Can be
float
or
double
.
pmethod
Not applicable
defaultDense
Available methods for multi-class classifier prediction stage:
  • defaultDense
    - the method described in [Wu04]
  • voteBased
    - the method based on the votes obtained from two-class classifiers.
tmethod
defaultDense
or
voteBased
training::oneAgainstOne
The computation method that was used to train the multi-class classifier model.
prediction
defaultDense
or
voteBased
Pointer to an object of the SVM prediction class
Pointer to the prediction algorithm of the two-class classifier. By default, the SVM two-class classifier is used.
nClasses
defaultDense
or
voteBased
Not applicable
The number of classes. A required parameter.
maxIterations
defaultDense
100
The maximal number of iterations for the algorithm.
accuracyThreshold
defaultDense
1.0e-12
The prediction accuracy.
resultsToEvaluate
voteBased
computeClassLabels
The 64-bit integer flag that specifies which extra characteristics of the decision function to compute.
Provide one of the following values to request a single characteristic or use bitwise OR to request a combination of the characteristics:
  • computeClassLabels
    for
    prediction
  • computeDecisionFunction
    for
    decisionFunction
Output
In addition to classifier output, multiclass classifier calculates the result described below. Pass the
Result ID
as a parameter to the methods that access the result of your algorithm. For more details, see Algorithms.
Result ID
Result
decisionFunction
A numeric table of size LaTex Math image. containing the results of the decision function computed for all binary models when the
computeDecisionFunction
option is enabled.
If
resultsToEvaluate
does not contain
computeDecisionFunction
, the result of
decisionFunction
table is
NULL
.
By default, each numeric table of this result is an object of the
HomogenNumericTable
class, but you can define the result as an object of any class derived from
NumericTable
except for
PackedSymmetricMatrix
and
PackedTriangularMatrix
.
Examples

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.