LinearClassifierBase<T>

Base Classes · AiDotNet.Classification.Linear

Provides a base implementation for linear classifiers.

Inheritance: ClassifierBase<T> → ProbabilisticClassifierBase<T> → LinearClassifierBase<T>

Implements: IGradientComputable<T, Matrix<T>, Vector<T>>

Known derived types: PassiveAggressiveClassifier<T>, PerceptronClassifier<T>, RidgeClassifier<T>, SGDClassifier<T>

For Beginners

Linear classifiers are one of the simplest forms of machine learning:

How they work:

  1. Each feature gets a weight (importance score)
  2. Multiply each feature by its weight and sum them up
  3. Add a bias term
  4. If the result is positive, predict one class; otherwise, the other

The training process adjusts the weights to correctly classify training examples.

Advantages:

  • Fast to train and predict
  • Work well with many features
  • Easy to interpret (weight = feature importance)
  • Often surprisingly effective

How It Works

Linear classifiers learn a linear decision function: f(x) = w * x + b where w is the weight vector and b is the bias (intercept).

Methods

Method Summary
IFullModel<T, Matrix<T>, Vector<T>> WithParameters(Vector<T> parameters)
Matrix<T> PredictLogProbabilities(Matrix<T> input)
Matrix<T> PredictProbabilities(Matrix<T> input)
ModelMetadata<T> GetModelMetadata()
Vector<T> ComputeGradients(Matrix<T> input, Vector<T> target, ILossFunction<T> lossFunction)
Vector<T> DecisionFunctionBatch(Matrix<T> input) Computes decision function values for all samples.
Vector<T> Predict(Matrix<T> input)
void ApplyGradients(Vector<T> gradients, T learningRate)