Smart Pass Mobile

• 

  A basic authentication screen with a twist

Passwords are not going anywhere. They are the gold standard for easy to implement, cost effective authentication, but they have downfalls.

Smart Pass Mobile is attempting to add a degree of intelligence to authenticating with a password. Instead of only looking at the string of characters, smart pass is looking at how the user behaves while inputing that password. This, though very much in research with our group, could one day be used as an effective, lightweight, and cost effective method of implicit two factor authentication.

Having a small group with limited knowledge on machine learning and data analytics. This is new territory for both me and my partner.

We were successful at keeping out about 70% of unauthorized of users out of this system entirely. The users that were able to authenticate to Smart Pass Mobile took several tries to get past the extra layer, even when given the password. In addition, Smart Pass Mobile was authenticating the intended users at significantly higher rates than impostors.

I learned a great deal about statistics modeling. I also learned about feature extraction, evidence pooling, and stylometric authentication.

Smart Pass Mobile has high potential. The algorithm that we used to derive our authentication data is extremely rudimentary. We chose this because of the time constraints of the competition. Given a larger amount of time for research and planning, I believe that smart pass can identify users as well as a finger print can. If this is possible, it could greatly simplify how we approach traditional pass-key authentication.

Built With

• android
• java
• statistics