Advanced Dialogue Systems and Converstational AI (DS-ConvAI-M)

This module deals with state-of-the-art approaches to Conversational AI - text-based or speech-based dialogue interaction through language - and its modelling and realisation through machine learning and deep learning. Building upon content of the module Introduction to Dialogue Systems, it dives into the technical realization of chatbots and spoken dialogue systems ranging from a modular pipeline architecture to end-to-end neural models including Large Language Models (LLMs). The module can be successfully completed without prior knowledge on dialogue systems.

In this course, students will learn/recap theoretical foundations about conversational AI and dialogue systems technology and modelling. Participants will learn about various technological aspects of conversational AI with a focus on state-of-the-art neural, and deep learning approaches to sequential and non-sequential supervised learning also touching the usage of linguistic representations such as word embeddings. Students will gain insights into dialogue modelling through reinforcement learning and deep reinforcement learning and how to derive a suitable objective function. Participants will learn how to make use of advanced deep learning architectures like recurrent neural networks and transformers for their application on various problems of dialogue systems and the dialogue system itself.

The lecture is accompanied by practicals and assignments that will help participants to develop practical, hands-on experience. In those practicals, students will implement and evaluate different approaches for dialogue systems and its modules using machine learning algorithms using Python and its respective commonly used libraries.

The following is a selection of topics that will be addressed in the course:

  • Machine-larning based methods to various spoken dialogue system modules
  • Statistical Spoken Dialogue Systems
  • Large Language Models and their application in Conversational AI
  • End-to-end Neural Dialogue Generators
  • Evaluation techniques


The lecture is accompanied by practicals and assignments and runs each summer term.

Participants generally need to be enrolled in one of the Master programms Applied Computer Science or Computing in the Humanities. Students enrolled in Information Systems should check with the Servicedesk WI if they are able to participate.

The course language is English.

Prerequisits: good working knowledge of programming (e.g., in Python)

Recommended (not mandatory) completion of modules: Einführung in die KI/Introduction to AI [AI-KI-B], Einführung in die Dialogsysteme/Introduction to Dialogue Systems [DS-IDS-M], Deep Learning [xAI-DL-M]