Paulius Dilkas

Research project title: 
Explainability in autonomous agents: interpretable models, abstraction, and beyond
Research project: 

My work revolves around programs and models that have both a relational (e.g., first-order logic) and a probabilistic structure, such as probabilistic logic programs and Markov logic networks. I investigate their fundamental properties as well as transformations and abstractions that can be used to reduce the computational cost of inference and/or improve the model's interpretability.

Student type: 
Current student