Alfeesi/cybersecurity-ml-models

🛡️ Cybersecurity ML Models

A collection of machine learning models for cybersecurity threat detection and vulnerability assessment.

Models Included

🔍 Intrusion Detection (UNSW-NB15)

Detects network intrusions and classifies attack types.

• Algorithm: Random Forest + XGBoost
• Classes: Normal, DoS, Exploits, Fuzzers, Generic, Reconnaissance, Backdoor, Analysis, Shellcode, Worms
• Dataset: UNSW-NB15

🎣 Phishing Detection

Identifies phishing URLs and malicious web content.

• Algorithm: Random Forest + XGBoost
• Output: Phishing / Legitimate

🛡️ Vulnerability Scoring

Predicts vulnerability severity and CVSS scores.

• Algorithm: Random Forest + XGBoost (classifier + regressor)
• Output: Severity label + numeric score

---

📁 Repository Structure

├── predictor.pkl                          # Main predictor (load this)
├── preprocessors/
│   ├── intrusion_label_encoder.pkl
│   ├── intrusion_scaler.pkl
│   ├── phishing_scaler.pkl
│   ├── severity_encoder.pkl
│   ├── vulnerability_scaler.pkl
│   ├── intrusion_feature_names.json
│   ├── phishing_feature_names.json
│   └── vulnerability_feature_names.json
└── saved_models/
    ├── intrusion_detection/
    ├── phishing_detection/
    └── vulnerability_scoring/

---

🚀 Quickstart

Install dependencies

pip install scikit-learn xgboost joblib huggingface_hub

Load the models

import joblib
from huggingface_hub import hf_hub_download

# Download and load main predictor
path = hf_hub_download(repo_id="Alfeesi/cybersecurity-ml-models", filename="predictor.pkl")
predictor = joblib.load(path)

Intrusion Detection

import numpy as np

features = np.array([[0.5, 0, 1, 1024, 512]])  # your network features
result = predictor.predict_intrusion(features)
print(f"Attack Type: {result['attack_type']}")

Phishing Detection

features = np.array([[75, 1, 3, 0]])  # url_length, has_ip, num_dots, has_https
result = predictor.predict_phishing(features)
print(f"Phishing: {result}")

Vulnerability Scoring

features = np.array([[7.5, 0, 0]])  # cvss_base, attack_vector, complexity
result = predictor.predict_vulnerability(features)
print(f"Severity: {result['severity']} | Score: {result['score']}")

---

📊 Model Performance

Model	Metric	Score
Intrusion Detection	Accuracy	See evaluation/
Phishing Detection	Accuracy	See evaluation/
Vulnerability Scoring	RMSE	See evaluation/

---

🔗 Live Demo

https://huggingface.co/spaces/Alfeesi/cybersecurity-demo

---

📄 License

Apache 2.0 — free to use, modify, and distribute.