Overview

The Keystroke Authentication System is a Machine Learning–based behavioral biometric application that authenticates users by analyzing their unique typing patterns. Instead of relying solely on passwords, the system captures keystroke dynamics such as dwell time (how long a key is pressed) and latency/flight time (time between consecutive key presses) to verify user identity and detect impersonation attempts.

Key Features

Behavioral biometric authentication using keystroke dynamics
Extraction of dwell time and latency features
Machine Learning–based user classification
Impersonation and unauthorized access detection
Secure login with User ID, Email ID, and Password
REST API integration using Flask

Technology Stack

Language: Python
Machine Learning: Scikit-learn
Backend: Flask
Database: PostgreSQL

How It Works

The user enters their User ID, Email ID, and Password.
The system records keystroke timing data while the user types.
Dwell time and latency features are extracted and processed.
A Scikit-learn model compares the typing behavior with the user's registered profile.
If the typing pattern matches, authentication is successful; otherwise, the system flags a potential impersonation attempt and denies access.

Applications

Enterprise employee authentication
Banking and financial systems
E-commerce platforms
Secure web applications
Identity verification systems

Outcome

This project demonstrates the use of behavioral biometrics and Machine Learning for secure user authentication. By combining Python, Scikit-learn, Flask, and PostgreSQL, the system enhances traditional password-based login with continuous behavioral verification, making authentication more secure and resistant to impersonation attacks.