KDnuggets Home » News » 2014 » Oct » News, Features » One-handed Keystroke Biometric Identification Competition ( 14:n26 )

One-handed Keystroke Biometric Identification Competition

Build a biometric keystroke classifier in this new competition to help identify the features that best predict one-handed typing samples. The prize for first place is a fingerprint scanner.

Pressing key One-handed Keystroke Biometric Identification Competition

Keystroke biometrics is the act of identifying or verifying an individual based on typing patterns. This competition addresses the situation in which the typing behavior of a user is seriously handicapped during the testing phase of a biometric system.

This situation may arise when a user has enrolled a normally-typed sample during the training phase and later is only able to type with one hand due to an injury or distraction during the testing phase.

A robust keystroke biometric application needs to be able to handle this situation gracefully. Is it possible to identify a one-handed typing sample? Should the user be re-enrolled while typing with only one hand? Or is some other fallback mechanism needed to attempt the identification of the individual? The results of this competition will help to answer that question.

The goal of the competition is to correctly classify the identity of as many unlabeled keystroke samples as possible. Classification results can submitted directly to the competition website where they will automatically be scored. Python starter code is provided to parse the data and quickly start building a model.

The dataset is very challenging. The benchmark score from the starter code (seen on the leaderboard page) shows classification accuracies of less than 5% for all conditions. This should be easy to beat, though, for reasons explained in the starter code.

Competition rules and prize:

Participants can make up to one submission per day, up to the competition deadline. The leaderboard will be calculated on 50% of the data until the last day of the competition to avoid overfitting.

The winner of the One-handed Keystroke Biometric Identification Competition will receive a Futronic FS88 Fingerprint Scanner!

For more information and instructions on how to participate, see the competition website: http://biometric-competitions.com/mod/competition/view.php?id=7