Skip to content

Godkunn/NeuroScan-AI-Security

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 

Repository files navigation

🛡️ NeuroScan: Robustness Analysis of Edge AI

The Problem: Medical AI needs to be portable (for phones) but also secure against errors. The Solution: I built a Quantized MobileNetV2 (73% smaller than standard) and tested it against adversarial attacks.

Key Findings:

Accuracy: 94% on Test Data.

Efficiency: Compressed from 9MB to 2.5MB.

Security Flaw: A noise injection of just 0.05 caused the AI to misdiagnose a tumor as healthy.

Code Structure:

NeuroScan: Vulnerability Analysis of Edge-Deployed Medical AI.ipynb: The training and attack experiments.

app.py: A live demo of the lightweight model.

About

Anyone can train an AI to detect brain tumors. I took it three steps further. I built a system that is 94% accurate, compressed it by 73% to run on a cheap phone, and then I hacked it to prove a critical security flaw in modern medical AI.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages