KrumFullModelAggregationStrategy<T, TInput, TOutput>

Models & Types · AiDotNet.FederatedLearning.Aggregators

Krum aggregation for IFullModel (Byzantine-robust selection by distance).

Inheritance: FederatedLearningComponentBase<T> → AggregationStrategyBase<IFullModel<T, TInput, TOutput>, T> → RobustFullModelAggregationStrategyBase<T, TInput, TOutput> → KrumFullModelAggregationStrategy<T, TInput, TOutput>

How It Works

For Beginners: Krum picks the single client update that is most consistent with the others. It does this by computing distances between client updates and selecting the one with the smallest sum of distances to its closest neighbors.

Constructors

Constructor Summary
KrumFullModelAggregationStrategy(int byzantineClientCount)

Methods

Method Summary
IFullModel<T, TInput, TOutput> Aggregate(Dictionary<int, IFullModel<T, TInput, TOutput>> clientModels, Dictionary<int, double> clientWeights)
string GetStrategyName()