RfaFullModelAggregationStrategy<T, TInput, TOutput>
Models & Types · AiDotNet.FederatedLearning.Aggregators
Robust Federated Aggregation (RFA) via geometric median (Weiszfeld iterations).
Inheritance: FederatedLearningComponentBase<T> → AggregationStrategyBase<IFullModel<T, TInput, TOutput>, T> → RobustFullModelAggregationStrategyBase<T, TInput, TOutput> → RfaFullModelAggregationStrategy<T, TInput, TOutput>
How It Works
For Beginners: Instead of averaging client updates, the geometric median finds a point that minimizes the sum of distances to all client updates. This is more robust when some clients are outliers or adversarial.
Constructors
| Constructor | Summary |
|---|---|
RfaFullModelAggregationStrategy(int maxIterations, double tolerance, double epsilon, bool useClientWeights) |
Methods
| Method | Summary |
|---|---|
IFullModel<T, TInput, TOutput> Aggregate(Dictionary<int, IFullModel<T, TInput, TOutput>> clientModels, Dictionary<int, double> clientWeights) |
|
RfaFullModelAggregationStrategy<T, TInput, TOutput> FromOptions(RobustAggregationOptions options) |
|
string GetStrategyName() |