Human-Machine Dialogue
Master in Artificial Intelligence Systems
Register on this course's Moodle page to receive updates on assignments, notices, and general information.
Description
Computers and agents engaging in conversations with humans are becoming increasingly prevalent across industries and applications. In this course, we present the fundamental principles and structures underlying human conversations. We draw on conversation linguistics to develop computational dialogue models, dialogue system architectures, and evaluation methodologies. The second part of the course equips students with methodologies for designing conversational agents, data-driven training (including generative AI models, agentic frameworks, and end-to-end pipelines), design tools, and a project-based lab that tackles real-world use cases. This approach ensures students ground concepts and learn to apply them in real-world scenarios.
Students will design, develop, and evaluate human-machine dialogue systems for an application with large language models and/or agentic frameworks.
Let Us Know
Use the student-lecturer space on didattica-online or
the feedback form (anonymous).
Lectures:
Course Presentation (Incl. Assignments and Student Evaluation)
Basics of Human-Computer Interaction
Linguistics of Conversations ( Part 1, Part 2 , Part 3 )
- Generating Spoken Conversations Lab
- Generating Emotions in Spoken Conversations Lab
Dialogue Models
Dialogue Evaluation
Reinforcement Learning and Applications to HMD
Post-training of LLMs (including RLHF)
Large Language Models ( Lecture and Lab)
Ethics and Conversational AI
LAB
Lab Workplan for the HMD project
Past project Reports : Spoken HMD with Alexa, Project with LLMs
