Network Intrusion Anomaly Detector

Anomaly-based detection learns what normal traffic looks like — and flags everything that deviates. Trained on 2.8M labeled network flows from CICIDS2017. Model runs entirely in your browser.

Try the Classifier →View Results
Approach
01
CICIDS2017
2.8M labeled flows
→
02
Feature Eng.
15 network features
→
03
Isolation Forest
Train on benign only
→
04
JSON Export
100 trees → ~2MB
→
05
Browser
TypeScript inference
Model Performance
F1 Score
—
Run notebook to populate
Attacks Detected
—
Run notebook to populate
False Positive Rate
—
Run notebook to populate