LeafFederatedSplit<TInput, TOutput>

Models & Types · AiDotNet.FederatedLearning.Benchmarks.Leaf

Represents a single LEAF split (train/test) as per-client datasets.

For Beginners

Instead of one big dataset for everyone, federated learning uses many small datasets — one per device or organization. This class stores that “one dataset per client” view of the data.

How It Works

LEAF is a federated learning benchmark suite where each user corresponds to one client. This type preserves that structure by storing one dataset per user.

Constructors

Constructor Summary
LeafFederatedSplit(IReadOnlyList<string> userIds, IReadOnlyDictionary<string, FederatedClientDataset<TInput, TOutput>> userData) Initializes a new instance of the LeafFederatedSplit class.

Properties

Property Summary
IReadOnlyDictionary<string, FederatedClientDataset<TInput, TOutput>> UserData { get; } Gets the per-user datasets in this split.
IReadOnlyList<string> UserIds { get; } Gets the ordered list of user IDs in this split.
int ClientCount { get; } Gets the number of clients/users in this split.

Methods

Method Summary
Dictionary<int, FederatedClientDataset<TInput, TOutput>> ToClientIdDictionary(IReadOnlyDictionary<int, string> clientIdToUserId) Converts this split into the Dictionary<int, FederatedClientDataset> form used by trainers.