qora-fl 0.2.0

Qora: Quorum-Oriented Robust Aggregation for Federated Learning

Qora-FL

Quorum-Oriented Robust Aggregation for Federated Learning

Byzantine-tolerant aggregation for federated learning. Handles up to 30% malicious clients with minimal overhead.

Features

• Trimmed Mean - Coordinate-wise outlier trimming (~30% Byzantine tolerance)
• Median - Coordinate-wise median (~50% Byzantine tolerance)
• Krum - Distance-based selection with fixed-point arithmetic (Blanchard et al., 2017)
• FedAvg - Standard baseline for comparison
• Reputation tracking - Automatically scores clients based on contribution quality
• Deterministic consensus - Optional I16F16 fixed-point math for bit-perfect reproducibility

Quick Start

cargo add qora-fl

use qora_fl::{ByzantineAggregator, AggregationMethod};
use ndarray::array;

let mut agg = ByzantineAggregator::new(AggregationMethod::TrimmedMean, 0.3);

let updates = vec![
    array![[1.0, 2.0]],   // Honest
    array![[1.1, 2.1]],   // Honest
    array![[0.9, 1.9]],   // Honest
    array![[100.0, 200.0]], // Byzantine attacker
];

let result = agg.aggregate(&updates, None).unwrap();
// Result is close to [1.0, 2.0], attacker ignored

Aggregation Methods

Method	Byzantine Tolerance	Use Case
TrimmedMean	~30% of clients	Default choice for most FL deployments
Median	~50% of clients	When stronger robustness is needed
Krum	n >= 2f+3	Fixed-point deterministic consensus
FedAvg	None	Baseline comparison only

Using Individual Functions

use qora_fl::{trimmed_mean, median, fedavg};
use ndarray::array;

let updates = vec![array![[1.0]], array![[2.0]], array![[3.0]]];

let tm = trimmed_mean(&updates, 0.2).unwrap();
let med = median(&updates).unwrap();
let avg = fedavg(&updates, None).unwrap();

Examples

Quickstart

cargo run --example quickstart

Compare Methods

cargo run --example compare_methods

See examples/ for more.

Background

Core algorithms validated in QRES (181-day autonomous IoT deployment with 30% Byzantine tolerance). Qora-FL adapts this proven consensus to standard federated learning workflows.

License

Licensed under either of Apache License, Version 2.0 or MIT License at your option.