BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:ML-based Text Steganography
DTSTART:20190709T113000
DTEND:20190709T133000
DTSTAMP:20260406T194607Z
UID:2b95b04b002537db00bd8dd5996ab3b34dae30291f15cc6972820694
CATEGORIES:Conferences - Seminars
DESCRIPTION:Andreas Hug\nEDIC candidacy exam\nExam president: Prof. Robert
  West\nThesis advisor: Prof. Katerina Argyraki\nThesis co-advisor: Prof. M
 artin Jaggi\nCo-examiner: Prof. Carmela Troncoso\n\nAbstract\nLinguistic s
 teganography systems are becoming more important with rising concerns and 
 awareness\nfor user privacy. The field of natural language processing has 
 recently witnessed tremendous\nimprovements in a multitude of applications
  that could potentially be useful.\nIn this report\, we discuss an existin
 g linguistic steganography approach\, a new encoder-decoder\narchitecture 
 and an instance of a generative adversarial network for discrete sequences
 . We discuss\nhow these three methods can be combined to create better lin
 guistic stegosystems. Lastly\, we give\na glimpse on the work done by us d
 uring this academic year as well as a short outlook on future\nwork.\n\nBa
 ckground papers\nMatryoshka: Hiding Secret Communication in Plain Sight\, 
 by Safaka\, I.\, Fragouli\, C.\, Argyraki K.\nAttention is all you need\, 
 by Vaswani\, A.\, et al.\nSeqGAN: Sequence Generative Adversarial Nets wit
 h Policy\, by Yu\, L.\, et al.\n 
LOCATION:BC 329 https://plan.epfl.ch/?room==BC%20329
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR