WinsorizedMeanFullModelAggregationStrategy<T, TInput, TOutput>
Models & Types · AiDotNet.FederatedLearning.Aggregators
Coordinate-wise winsorized mean aggregation for IFullModel.
Inheritance: FederatedLearningComponentBase<T> → AggregationStrategyBase<IFullModel<T, TInput, TOutput>, T> → RobustFullModelAggregationStrategyBase<T, TInput, TOutput> → WinsorizedMeanFullModelAggregationStrategy<T, TInput, TOutput>
How It Works
For Beginners: Winsorized mean is like trimmed mean, but instead of dropping extreme values, it clips them to the nearest remaining value before averaging. This reduces the impact of outliers while keeping the same number of values in the average.
Constructors
| Constructor | Summary |
|---|---|
WinsorizedMeanFullModelAggregationStrategy(double winsorizeFraction) |
Methods
| Method | Summary |
|---|---|
IFullModel<T, TInput, TOutput> Aggregate(Dictionary<int, IFullModel<T, TInput, TOutput>> clientModels, Dictionary<int, double> clientWeights) |
|
string GetStrategyName() |