Can mobile phone users be identified by movement?

July 26, 2013 - PRLog -- Since everyone moves differently and accelerometers are fast becoming ubiquitous, the contest aims to investigate the feasibility of using accelerometer data as a biometric for identifying users of mobile devices.

Seal has collected accelerometer data from several hundred users over a period of several months, using the Quickdraw app (https://play.google.com/store/apps/details?id=com.seal.quickdraw&feature=search_result#?t=W251bGwsMSwxLDEsImNvbS5zZWFsLnF1aWNrZHJhdyJd) published on Googles’s Android PlayStore. The app samples accelerometer data in the background  during normal device operation and posts the data to a central database.

60 million unique samples of accelerometer data are available to contestants for analysis. The data is split into a training set labeled with the device that the sample was collected from, and a test set demarcated into sequences of 300 samples. To compete for the $5000  in prize money, contestants predict the likelihood that a sequence of samples presented in the test set was generated from a declared device. The competition runs until November, 22 2013.

Seal (http://www.seal-id.com) was founded in 2012 to research and provide solutions for enabling unobtrusive authentication of mobile device users.

Kaggle (http://www.kaggle.com/competitions) is the leading platform for predictive modeling competitions. With 100,000 data scientists as registered members, Kaggle is the largest and most active data science community on or off the web.