ModelUpdateVerifier<T>
Models & Types · AiDotNet.FederatedLearning.Verification
Server-side verification engine that checks all proofs from clients before aggregation.
Inheritance: FederatedLearningComponentBase<T> → ModelUpdateVerifier<T>
For Beginners
The model update verifier is the server-side component that examines cryptographic proofs from each client before their updates are included in the global model. Think of it as a “bouncer” that checks credentials before letting clients contribute.
How It Works
Verification workflow:
Constructors
| Constructor | Summary |
|---|---|
ModelUpdateVerifier(VerificationOptions options) |
Initializes a new instance of ModelUpdateVerifier. |
Properties
| Property | Summary |
|---|---|
IGradientCommitment<T> CommitmentScheme { get; } |
Gets the commitment scheme for clients to use. |
IReadOnlyList<VerificationRecord> VerificationHistory { get; } |
Gets the verification history for auditing. |
Methods
| Method | Summary |
|---|---|
ClientVerificationResult VerifyClientUpdate(int clientId, int round, GradientCommitmentData<T> commitment, VerificationProof normProof, VerificationProof boundednessProof, VerificationProof lossProof) |
Verifies a client’s update including commitment and any required proofs. |
int GetRejectedClientCount(int round) |
Gets the count of clients that failed verification in a given round. |
int GetVerifiedClientCount(int round) |
Gets the count of clients that passed verification in a given round. |