ruqu-qflg

Quantum Federated Learning with Byzantine Tolerance - Privacy-preserving distributed quantum ML.

Part of the ruQu quantum computing suite by ruv.io.

Features

• Federated Aggregation - Secure gradient aggregation across clients
• Byzantine Tolerance - Robust to malicious or faulty participants
• Differential Privacy - Formal privacy guarantees with ε-δ bounds
• Quantum Secure - Post-quantum cryptographic primitives
• Async Communication - Non-blocking client updates

Installation

[dependencies]
ruqu-qflg = "0.1"

Quick Start

use ruqu_qflg::{FederationServer, ClientConfig, PrivacyConfig};

let privacy = PrivacyConfig {
    epsilon: 1.0,
    delta: 1e-5,
    clip_norm: 1.0,
};

let server = FederationServer::new(10, privacy)?;

// Federated training round
let gradients = server.collect_gradients().await?;
let aggregated = server.aggregate_byzantine_robust(&gradients)?;
server.broadcast_update(&aggregated).await?;

License

MIT License - see LICENSE