ssam18

ssam18 / zta-federated-learning

Public

Zero-Trust Agentic Federated Learning for Secure IIoT Defense Systems (arXiv:2512.23809)

19
2
100% credibility
Found Apr 19, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A research toolkit for testing privacy-safe AI teamwork to detect cyber threats in industrial factories using real-world data.

How It Works

1
🔍 Discover Secure Factory Guardian

You learn about a clever project that trains smart detectors to spot hackers in factories without anyone sharing private info.

2
📥 Grab the Toolkit

Download the ready-to-use files to your computer in moments.

3
🛠️ Prep Your Playground

Follow easy guides to set up your space – everything clicks into place quickly.

4
🚀 Kick Off the Team Training

Start the magic: watch AI helpers from different factories team up safely to learn spotting dangers.

5
📈 Watch Results Unfold

Numbers and colorful charts pop up, proving your guardians handle tricks and bad actors best.

🏆 Victory: Ironclad Protection

Celebrate! You've tested unbeatable AI shields keeping factories safe from cyber sneaks.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is zta-federated-learning?

This Python framework delivers zero-trust agentic federated learning for IIoT intrusion detection, letting edge devices collaboratively train CNN-LSTM models without sharing raw network traffic data. It tackles vulnerabilities like Byzantine poisoning, adversarial evasion, and Sybil attacks via TPM attestation, robust aggregation, and on-device adversarial training in an edge-fog-cloud setup. Users get preprocessed public datasets (Edge-IIoTset, CIC-IDS2017, UNSW-NB15), CLI scripts to run baselines, ablations, and robustness tests, plus JSON results for plotting paper-grade figures.

Why is it gaining traction?

It bundles end-to-end repro of arXiv:2512.23809 results, beating FedAvg/Krum/FLAME on clean accuracy (97.8% on Edge-IIoTset) and under 30% Byzantine attacks (93% acc). The agentic zero trust protocol shines with SHAP-weighted updates and attestation, making FL secure for real IIoT without custom plumbing. Devs dig the quick smoke tests and GPU flags for fast iteration.

Who should use this?

IIoT security engineers hardening anomaly detectors across PLCs/SCADA sites. Researchers benchmarking zero-trust federated learning against poisoning/adversarial threats. Defense system devs prototyping agentic AI zero trust architectures on heterogeneous edge fleets.

Verdict

Grab it for IIoT FL experiments—docs, notebooks, and pytest coverage make it runnable out-of-box, reproducing strong results despite 19 stars and 1.0% credibility signaling early maturity. Test on your data before production.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.