OnlineSGDClassifier<T>

Models & Types · AiDotNet.OnlineLearning

Online Stochastic Gradient Descent classifier for incremental learning.

Inheritance: OnlineLearningModelBase<T> → OnlineSGDClassifier<T>

For Beginners

This classifier learns to separate two classes using a linear decision boundary, updating itself one example at a time.

How it works:

  1. Compute prediction: P(y=1) = sigmoid(w·x + b)
  2. Compare with true label: error = prediction - truth
  3. Update weights: w = w - learning_rate × error × x

The model “nudges” itself toward correct predictions with each example. Over many examples, it converges to a good decision boundary.

Advantages:

  • Handles streaming data naturally
  • Memory-efficient (doesn’t store data)
  • Can adapt to changing patterns

Supports:

  • L1 regularization (sparse weights)
  • L2 regularization (smooth weights)
  • Elastic Net (combination)

Usage:

References:

  • Bottou, L. (2010). “Large-Scale Machine Learning with Stochastic Gradient Descent”

How It Works

Online SGD Classifier implements logistic regression with SGD updates, allowing the model to learn incrementally from streaming data.

Constructors

Constructor Summary
OnlineSGDClassifier(double learningRate, LearningRateSchedule learningRateSchedule, double l1Penalty, double l2Penalty, bool fitIntercept) Gets the model type.

Methods

Method Summary
Dictionary<string, T> GetFeatureImportance() Gets the feature importance scores (absolute weights).
IFullModel<T, Matrix<T>, Vector<T>> WithParameters(Vector<T> parameters) Creates a new instance with specified parameters.
T DecisionFunction(Vector<T> x) Computes the score used for decision (before sigmoid).
T GetBias() Gets the bias (intercept) term.
T PredictProbability(Vector<T> x) Predicts the probability of class 1.
T PredictSingle(Vector<T> x) Predicts the target value for a single sample.
Vector<T> GetWeights() Gets the weights vector.
Vector<T> PredictProbabilities(Matrix<T> x) Predicts probabilities for all samples.
void PartialFit(Vector<T> x, T y) Updates the model with a single training example.
void Reset() Resets the model to its initial state.