Jaggi, MartinLagarde, Marie-Jeanne2018-06-252018-06-252018-06-252018-06-21https://infoscience.epfl.ch/handle/20.500.14299/146972The aim of this project is to assess the non- detectability of slightly modified version of the linguistic steganography approach described in the paper Generating Steganographic Text with LSTMs by classifying whether a tweet was generated by a regular user or by this stegano- graphic method. For that purpose, both a fastText supervised classification and a user study are conducted.SteganographyDetectionLSTMstudent work::semester or other student projects