IFederatedDriftDetector<T>

Interfaces · AiDotNet.FederatedLearning.DriftDetection

Detects concept drift in federated learning by monitoring client model updates over time.

Known implementations: ModelDriftDetector<T>, StatisticalDriftDetector<T>

For Beginners

In production FL, data doesn’t stay static. Customer behavior changes, new fraud patterns emerge, market conditions shift. This interface monitors each client’s training behavior and flags when their data distribution has changed significantly. The system can then adapt (e.g., increase that client’s training epochs, adjust aggregation weights, or trigger selective retraining).

How It Works

Usage:

Properties

Property Summary
string MethodName { get; } Gets the name of the drift detection method.

Methods

Method Summary
Dictionary<int, double> GetAdaptiveWeights(Dictionary<int, double> originalWeights, DriftReport driftReport) Gets recommended aggregation weight adjustments based on detected drift.
DriftReport DetectDrift(int round, Dictionary<int, Tensor<T>> clientModels, Tensor<T> globalModel, Dictionary<int, double> clientMetrics) Detects drift across all federated clients for the current round.
void Reset() Resets drift detection state for all clients.