Developed a system for analysing keystroke (Keystroke Dynamics) to authenticate users based on their inherent typing behaviour.
We have made various models using SVM, Manahattan Detector and Neural Networks. Also developed a basic GUI for generating and retreiving keystroke data.
Our Neural Network is reaching upto 91% accuracy on the test set.
Link for the dataset.
https://www.cs.cmu.edu/~keystroke/
This is a very comprehensive source for anyone who wants to try out keystroke analysis.
https://www.cs.cmu.edu/~keystroke/KillourhyMaxion09.pdf
The above research paper compares various anomaly detection algorithms and how they fare for this purpose.