Biometrics: Identification in Action
As part of a collaboration with Google Inc., with the University of Guelph, Canada, and with MIT, LIRIS designed an system for the automatic authentication of smartphone users.
The work was carried out by Natalia Neverova as part of her Phd thesis at LIRIS and INSA Lyon (thesis defended in April 2016). During her exchange stay in Mountain View USA, she developed a new deep recurrent network specifically designed for this task, which recognizes a user by his/her movements (linear and angular accelerations) measured by the integrated sensor of a smartphone.
more information :
Scientific article :
Natalia Neverova, Christian Wolf, Lacey Griffin, Lex Fridman, Deepak Chandra, Brandon Barbello & Graham W. Taylor (2016). « Learning Human Identity from Motion Patterns ». IEEE Access, . ArXiv : 1511.03908. HAL : hal-01281946.