VflUnlearningMethod
Enums · AiDotNet.Models.Options
Specifies the method used to remove an entity’s influence from a trained VFL model.
For Beginners
Under GDPR and similar regulations, individuals have the “right to be forgotten”. If a patient asks to be removed from a model, the model must be updated so that it no longer contains any information learned from that patient’s data. These methods vary in how thoroughly they remove the influence and how computationally expensive they are.
Fields
| Field | Summary |
|---|---|
VflUnlearningMethod Certified |
Certified unlearning with mathematical guarantees that the unlearned model is statistically indistinguishable from a model trained without the removed data. |
VflUnlearningMethod GradientAscent |
Apply gradient ascent on the removed entities to approximately reverse their influence. |
VflUnlearningMethod PrimalDual |
Uses a primal-dual optimization framework for both sample and label unlearning in VFL. |
VflUnlearningMethod Retraining |
Retrain the model from scratch without the removed entities. |
int value__ |