Machine Reasoning
DACS Research - Artificial Intelligence
Our research in machine reasoning focuses on combining learned knowledge to identify potential follow-up actions and their consequences. We advance search and planning techniques for decision-making, enabling AI to make informed, strategic choices in complex environments such as manufacturing and games. We also develop formal methods to mitigate bias in classification models, enhancing fairness and trustworthiness. Additionally, our work on mechanistic interpretation and optimization of foundation models improves their reliability.