FederatedScoreAggregation
Enums · AiDotNet.Enums
How a federated feature selector combines the per-client feature scores into one global score.
For Beginners
Each client scores the features on its own data. The aggregation decides how the clients’ opinions are merged: an average, an average weighted by how much data each client has, or a robust choice (median, trimmed mean) that ignores clients whose scores look extreme.
Fields
| Field | Summary |
|---|---|
FederatedScoreAggregation Mean |
The plain average of the client scores. |
FederatedScoreAggregation Median |
The median of the client scores (the mean of the two middle values for an even client count). |
FederatedScoreAggregation TrimmedMean |
The mean after dropping the max(1, clients / 5) lowest and highest client scores. |
FederatedScoreAggregation WeightedAverage |
Average weighted by each client’s number of samples. |
int value__ |