VerticalFederatedLearningOptions

Options & Configuration · AiDotNet.Models.Options

Top-level configuration for vertical federated learning (VFL).

For Beginners

Vertical Federated Learning allows multiple parties that hold different features for the same entities to jointly train a model without sharing their raw data. For example, a bank (income, credit score) and a hospital (diagnoses, prescriptions) can jointly predict loan default risk.

How It Works

This is fundamentally different from horizontal FL (where each party has the same features for different samples). VFL requires:

Example:

Constructors

Constructor Summary
VerticalFederatedLearningOptions()

Properties

Property Summary
MissingFeatureOptions MissingFeatures { get; set; } Gets or sets options for handling missing features across parties.
PsiOptions EntityAlignment { get; set; } Gets or sets options for Private Set Intersection used to align entities across parties.
SplitModelOptions SplitModel { get; set; } Gets or sets options for the split neural network architecture.
VflUnlearningOptions Unlearning { get; set; } Gets or sets options for GDPR-compliant entity unlearning.
bool EnableLabelDifferentialPrivacy { get; set; } Gets or sets whether to apply differential privacy noise to the label holder’s gradients.
bool EncryptGradients { get; set; } Gets or sets whether to encrypt gradients exchanged between parties.
bool VerboseLogging { get; set; } Gets or sets whether to log detailed training metrics (loss, alignment stats, etc.) at each epoch.
double LabelDpDelta { get; set; } Gets or sets the differential privacy delta for label protection.
double LabelDpEpsilon { get; set; } Gets or sets the differential privacy epsilon for label protection.
double LearningRate { get; set; } Gets or sets the learning rate for model training.
int BatchSize { get; set; } Gets or sets the batch size for mini-batch training.
int NumberOfEpochs { get; set; } Gets or sets the number of training epochs.
int NumberOfParties { get; set; } Gets or sets the number of parties participating in VFL.
int? RandomSeed { get; set; } Gets or sets the random seed for reproducible training.