Design of algorithms with mathematical guarantees with respect to requirements concerning fairness of decisions, maintaining privacy of user data, robustness in presence of noisy data, and safe operation.
Tools for verifying that the systems with AI-based components satisfy their formal requirements.
Techniques focused on explaining decisions to stakeholders.
Integrating data-driven techniques such as deep learning with symbolic and logical reasoning techniques.
Design principles to tightly integrate humans in all stages of the machine-learning pipeline.
Exploring opportunities in the broad areas of health and medicine in collaboration with clinicians and medical researchers.