sumedh-r-m-6328/neurometric_guard_model

🛡️ NeuroMetric Guard: Intruder Detection Model

This is the trained Random Forest Classifier used in the NeuroMetric Guard Desktop Application. It detects unauthorized users based on micro-kinematic mouse movement patterns.

Model Details

• Architecture: Random Forest Classifier (n_estimators=300)
• Input Features: 10 (Speed, Jerk, Path Efficiency, Spectral Energy, Entropy, etc.)
• Classes:
  • 0: Authorized Owner
  • 1: Intruder
• Performance: 74% F1-Score (Single-session validation)

Intended Use

This model is designed to be run locally via the NeuroMetric Guard Python backend. It requires a window of 60 mouse events (approx. 0.5s) to generate a prediction.

Limitations

• Trained on a specific hardware sensitivity (DPI).
• Performance may degrade if the user switches mouse hardware without retraining.

How to Load

import joblib
from huggingface_hub import hf_hub_download

model_path = hf_hub_download(repo_id="sumedh-r-m-6328/neurometric-guard-model", filename="intruder_detector_model(72).pkl")
model = joblib.load(model_path)