Conversational interfaces based on deep learning are becoming more and
more ubiquitous. However, the massive amounts of stored speech and
text data that is needed for training state-of-the-art models raises
serious privacy concerns for its users. Each spoken message may
potentially reveal information about the user’s personality, may
contain critical information (credit card numbers, passwords, etc.),
and may convey sensitive information (ethnicity, age, health status,
etc.). Voice recordings may even be malevolently used to build
synthesized voiced to impersonate users.
The Spoken Language Systems group at Saarland University is seeking
new ways to provide dialog technology that is “private by design” by
means such as e.g. privacy-preserving machine learning. To this end,
we are anticipating the availability of a PhD position starting at the
beginning of 2019.
Ideal candidates for the position would have:
1. A good understanding of NLP and dialog phenomena
2. Excellent knowledge of machine learning
3. Excellent programming skills
4. Experience with architecting large systems.
About Saarland University
Saarland University is a modern university within the Saar-Lor-Lux region – a dynamic part of Europe defined by the shared borders between Germany, France and Luxembourg. An international perspective has been a defining feature of Saarland University ever since it was established in cooperation with France in 1948. Today, 15,000 young people are studying in Saarbrücken and Homburg (Faculty of Medicine), 17 per cent of whom are international students.