ContributionMethod
Enums · AiDotNet.Models.Options
Specifies the method used to evaluate client contributions in federated learning.
For Beginners
Not all clients contribute equally to the global model. Some provide high-quality data that greatly improves accuracy, while others may free-ride. These methods measure each client’s contribution so you can fairly compensate them or detect problems.
Fields
| Field | Summary |
|---|---|
ContributionMethod DataShapley |
Data Shapley: Monte Carlo approximation of Shapley values. |
ContributionMethod LightweightShapley |
Lightweight Shapley: efficient approximation using gradient similarity instead of full retraining. |
ContributionMethod Prototypical |
Prototypical contribution: evaluates contribution using prototype representations. |
ContributionMethod ShapleyValue |
Exact Shapley value: measures marginal contribution by testing all possible coalitions. |
int value__ |