In this seminar, we will discuss different dimensions of reliability of AI. We will study a diverse set of tasks and approaches, for example: Robustness, Selective Prediction, Modularity, Trustworthiness of AI, Uncertainty estimation, Evaluation of Reliability, or Out-of-Distribution Modeling.

The main focus of this seminar changes each semester. After successfully completing the seminar, students will be familiar with ongoing research in Reliable AI. Among other things, the seminar may cover:

• Basics of scientific presentations and reviewing
• Independent familiarization with current publications in Reliable AI
• Presentation of an existing publication
• Writing a scientific “mock” review of another publication
• Guiding the interactive discussion after the presentation
• Active participation in discussions, including feedback to presenters