M.Sc. Sabine Wieluch

Research

Research Interests

My research interest in Spoken Dialogue Systems (SDS) lies in learned sequence extraction from spoken text. I want to enable systems to learn certain (task) sequences from natural, freely spoken text to support future users. Therefore, a system is trained on a number of spoken sequence descriptions, using MachineLearning. As these descriptions will vary in description detail, length and even task order (some tasks might be interchangeable or the order does not matter), it is important to learn over a huge data set. Hence, an iterative learning approach is useful as the system can support the user while still in the learning process. Using Active Learning strategies, the system is able to pro-actively ask the user in uncertain situations and update the knowledge base. Finally, the knowledge base might still contain inconsistencies and conflicts. I’m interested in how these conflicts can be presented to the user without loosing trust in the system or displeasing him.

  • Active Learning
  • Machine Learning
  • Deep understanding of spoken text

Teaching

Summer Semester 2017:

  • Einführung in die Informatik
  • Repetitorium EidI
  • Programmierstarthilfe
  • Seminar Research Trends in Dialogue Systems

Research Assistant