InMemoryFederatedTrainer<T, TInput, TOutput>

Models & Types · AiDotNet.FederatedLearning.Trainers

In-memory federated learning trainer for local simulation and tests.

Inheritance: FederatedLearningComponentBase<T> → FederatedTrainerBase<IFullModel<T, TInput, TOutput>, FederatedClientDataset<TInput, TOutput>, FederatedLearningMetadata, T> → InMemoryFederatedTrainer<T, TInput, TOutput>

For Beginners

InMemoryFederatedTrainer provides AI safety functionality. Default values follow the original paper settings.

Constructors

Constructor Summary
InMemoryFederatedTrainer(IOptimizer<T, TInput, TOutput> optimizerPrototype, double? learningRateOverride, int? randomSeed, double convergenceThreshold, int minRoundsBeforeConvergence, FederatedLearningOptions federatedLearningOptions, IPrivacyMechanism<Vector<T>> differentialPrivacyMechanism, IPrivacyAccountant privacyAccountant, IClientSelectionStrategy clientSelectionStrategy, IFederatedServerOptimizer<T> serverOptimizer, IFederatedHeterogeneityCorrection<T> heterogeneityCorrection, IHomomorphicEncryptionProvider<T> homomorphicEncryptionProvider, FederatedAdvancedCapabilities<T> advancedCapabilities, ILogger logger)

Methods

Method Summary
FederatedLearningMetadata Train(Dictionary<int, FederatedClientDataset<TInput, TOutput>> clientData, int rounds, double clientSelectionFraction, int localEpochs)
FederatedLearningMetadata TrainRound(Dictionary<int, FederatedClientDataset<TInput, TOutput>> clientData, double clientSelectionFraction, int localEpochs)
ValueTuple<Vector<T>, UnlearningCertificate>? UnlearnClient(int clientId) Removes a client’s contribution from the trained global model (GDPR right-to-be-forgotten), when an unlearner was configured.