Table of Contents

Class OneVsRestClassifier<T>

Namespace
AiDotNet.Classification.Meta
Assembly
AiDotNet.dll

One-vs-Rest (also called One-vs-All) classifier for multi-class classification.

public class OneVsRestClassifier<T> : MetaClassifierBase<T>, IProbabilisticClassifier<T>, IMultiLabelClassifier<T>, IClassifier<T>, IFullModel<T, Matrix<T>, Vector<T>>, IModel<Matrix<T>, Vector<T>, ModelMetadata<T>>, IModelSerializer, ICheckpointableModel, IParameterizable<T, Matrix<T>, Vector<T>>, IFeatureAware, IFeatureImportance<T>, ICloneable<IFullModel<T, Matrix<T>, Vector<T>>>, IGradientComputable<T, Matrix<T>, Vector<T>>, IJitCompilable<T>

Type Parameters

T

The numeric data type used for calculations.

Inheritance
OneVsRestClassifier<T>
Implements
IFullModel<T, Matrix<T>, Vector<T>>
IModel<Matrix<T>, Vector<T>, ModelMetadata<T>>
IParameterizable<T, Matrix<T>, Vector<T>>
ICloneable<IFullModel<T, Matrix<T>, Vector<T>>>
IGradientComputable<T, Matrix<T>, Vector<T>>
Inherited Members
Extension Methods

Remarks

Trains one binary classifier per class, treating it as the positive class and all other classes as the negative class.

For Beginners: One-vs-Rest is a simple strategy for multi-class classification:

For 3 classes (A, B, C):

  • Classifier 1: Is it A vs not-A?
  • Classifier 2: Is it B vs not-B?
  • Classifier 3: Is it C vs not-C?

For prediction, the class whose classifier gives the highest score wins.

Advantages:

  • Simple and effective
  • Trains K classifiers for K classes
  • Easily parallelizable

Disadvantages:

  • Class imbalance (one class vs all others)
  • Classifiers don't see inter-class relationships

Constructors

OneVsRestClassifier(Func<IClassifier<T>>, MetaClassifierOptions<T>?, IRegularization<T, Matrix<T>, Vector<T>>?)

Initializes a new instance of the OneVsRestClassifier class.

public OneVsRestClassifier(Func<IClassifier<T>> estimatorFactory, MetaClassifierOptions<T>? options = null, IRegularization<T, Matrix<T>, Vector<T>>? regularization = null)

Parameters

estimatorFactory Func<IClassifier<T>>

Factory function to create base binary classifiers.

options MetaClassifierOptions<T>

Configuration options for the classifier.

regularization IRegularization<T, Matrix<T>, Vector<T>>

Optional regularization strategy.

Properties

NumLabels

Gets the number of labels that can be predicted.

public int NumLabels { get; }

Property Value

int

Methods

Clone()

Creates a clone of the classifier model.

public override IFullModel<T, Matrix<T>, Vector<T>> Clone()

Returns

IFullModel<T, Matrix<T>, Vector<T>>

A new instance of the model with the same parameters and options.

CreateNewInstance()

Creates a new instance of the same type as this classifier.

protected override IFullModel<T, Matrix<T>, Vector<T>> CreateNewInstance()

Returns

IFullModel<T, Matrix<T>, Vector<T>>

A new instance of the same classifier type.

GetModelType()

Returns the model type identifier for this classifier.

protected override ModelType GetModelType()

Returns

ModelType

Predict(Matrix<T>)

Predicts class labels for the given input data by taking the argmax of probabilities.

public override Vector<T> Predict(Matrix<T> input)

Parameters

input Matrix<T>

The input features matrix where each row is an example and each column is a feature.

Returns

Vector<T>

A vector of predicted class indices for each input example.

Remarks

This implementation uses the argmax of the probability distribution to determine the predicted class. For binary classification with a custom decision threshold, you may want to use PredictProbabilities() directly and apply your own threshold.

For Beginners: This method picks the class with the highest probability for each sample.

For example, if the probabilities are [0.1, 0.7, 0.2] for classes [A, B, C], this method returns class B (index 1) because it has the highest probability (0.7).

PredictLogProbabilities(Matrix<T>)

Predicts log-probabilities for each class.

public override Matrix<T> PredictLogProbabilities(Matrix<T> input)

Parameters

input Matrix<T>

The input features matrix where each row is a sample and each column is a feature.

Returns

Matrix<T>

A matrix where each row corresponds to an input sample and each column corresponds to a class. The values are the natural logarithm of the class probabilities.

Remarks

The default implementation computes log(PredictProbabilities(input)). Subclasses that compute log-probabilities directly (like Naive Bayes) should override this method for better numerical stability.

For Beginners: Log-probabilities are probabilities transformed by the natural logarithm. They're useful for numerical stability when working with very small probabilities.

For example:

  • Probability 0.9 → Log-probability -0.105
  • Probability 0.1 → Log-probability -2.303
  • Probability 0.001 → Log-probability -6.908

Log-probabilities are always negative (since probabilities are between 0 and 1). Higher (less negative) values mean higher probability.

PredictMultiLabel(Matrix<T>)

Predicts binary indicators for each label for each sample.

public Matrix<T> PredictMultiLabel(Matrix<T> input)

Parameters

input Matrix<T>

The input feature matrix.

Returns

Matrix<T>

A binary matrix where each row is a sample and each column is a label indicator (1=present, 0=absent).

PredictMultiLabelProbabilities(Matrix<T>)

Predicts probabilities for each label for each sample.

public Matrix<T> PredictMultiLabelProbabilities(Matrix<T> input)

Parameters

input Matrix<T>

The input feature matrix.

Returns

Matrix<T>

A probability matrix where each row is a sample and each column is the probability of that label.

PredictProbabilities(Matrix<T>)

Predicts class probabilities for each sample in the input.

public override Matrix<T> PredictProbabilities(Matrix<T> input)

Parameters

input Matrix<T>

The input features matrix where each row is a sample and each column is a feature.

Returns

Matrix<T>

A matrix where each row corresponds to an input sample and each column corresponds to a class. The values represent the probability of the sample belonging to each class.

Remarks

This abstract method must be implemented by derived classes to compute class probabilities. The output matrix should have shape [num_samples, num_classes], and each row should sum to 1.0.

For Beginners: This method computes the probability of each sample belonging to each class. Each row in the output represents one sample, and each column represents one class. The values in each row sum to 1.0 (100% total probability).

Train(Matrix<T>, Vector<T>)

Trains the One-vs-Rest classifier on the provided data.

public override void Train(Matrix<T> x, Vector<T> y)

Parameters

x Matrix<T>
y Vector<T>
Edit this page