ExactRetrainingUnlearner<T>
Models & Types · AiDotNet.FederatedLearning.Unlearning
Gold-standard unlearning: retrains the model from scratch excluding the target client.
Inheritance: FederatedLearningComponentBase<T> → ExactRetrainingUnlearner<T>
Implements: IFederatedUnlearner<T>
For Beginners
The most reliable way to forget a client is to pretend they never existed and retrain everything from the beginning. This is like erasing someone’s name from every page of a book and rewriting the book — it’s perfect but very slow.
How It Works
How it works:
When to use: Regulatory audits requiring provable unlearning, or when approximate methods fail verification. Very expensive — O(R * C) where R = rounds, C = clients.
Constructors
| Constructor | Summary |
|---|---|
ExactRetrainingUnlearner(FederatedUnlearningOptions options) |
Initializes a new instance of ExactRetrainingUnlearner. |
Properties
| Property | Summary |
|---|---|
string MethodName { get; } |
Methods
| Method | Summary |
|---|---|
ValueTuple<Tensor<T>, UnlearningCertificate> Unlearn(int targetClientId, Tensor<T> globalModel, Dictionary<int, List<Tensor<T>>> clientHistories) |