Keyboard Dynamics - A better way to authenticate

Download as .zip Download as .tar.gz View on GitHub Home

Typer-s (s for secure)

(Still under Research and Development - expecting completion by April '16)

A software piece written in python which is implemented on top of AI through ML to make your environment secure. I believe every one who is belonging to homo-sapiens has a unique pattern. It was difficult to implement an algorithm to analyze this complex pattern earlier. But as the advancement of technology we could make our path easier to the success. I just tried to use one logic, you may find a better one. Please don't hesitate to mail me that. Have a great typing pattern.

KeyBoard Dynamics??

Keyboard dynamics is a concept came and believed as an equivalent authentication hardening mechanism as biometrics. Its using the pattern analysis for security when people thought patterns can be unique. In fact, the concept was right to an extent. But by introducing the notion of AI and broadly implementing ML up on it, we would be able to achieve highly reliable security that might be stronger than biometrics.


This package is just a 1.5 month hard work in which I have managed to incorporate the different timely parameters of each key and create a huge db tree with that info in OOPS. I could also include the different pattern deciding erroneous factors such as backspacing, commonly used words etc. into it. I am sure if you try hard enough to find the patterns that make you stand out from others, you can have an awesome signature which you will generate through script and the script will compare always in your system without you knowing any complexities. You can add this to your servers, home system, or office system or wherever you want which has a keyboard to type. You can extent the feature of analyzing the mouse clicking and moving pattern and you can add your usual activity analysis and lot more, who is stopping you. Fork my git and give me 'pull' that's it right?!!

What is it doing!!

I could implement different algorithm which I have created by modifying existing ones as per my requirement. Important among these are Tree creation and data retrieval. Each key stroke is considered as an object that has different parameters. These parameters decides the pattern of the user and store in the tree safely. Script would be analyzing the pattern always and eventually algorithm is learning the pattern always. When it could find an anomaly, it will raise notification to the actual user or lock the machine or wipe the data or .. or .. Script should run in learning mode till it get enough data for comparing. We have '-l' as the switch to run the script in learning mode during which the bot would not compare the user pattern. It just LEARN!!

Authors and Contributors

In 2016, typer-s is copy lefted to @hhsecond

Support or Contact

Having trouble, ping me :)
mail me: hhsecond@outlook.com
tweet: @hhsecond