Foundations of Systems AI
The Research Department Foundations of Systems AI (SAINT) develops the algorithmic and programmatic foundations of Systems AI. Akin to systems biology, SAINT takes a systems view of AI that captures, understands, and leverages the interplay of different AI algorithms as building blocks of a single, complex AI system in a mathematically and computationally sound way.
“The focus of my research area's work is on how different AI building blocks can be synergistically combined: Low-level perception and high-level reasoning, which are typically treated with different approaches. While low-level perception, such as recognizing a road sign, can be handled very well with Deep Learning nowadays, high-level reasoning, such as concluding that a road sign at 120 km/h makes no sense in a small village, is captured by using logical and probabilistic representations. To address the grand challenge, we are developing the algorithmic foundations for “Systemic AI”, where interactions of individual AI building blocks (algorithms) are captured, understood, and used to describe a single, complex AI system. The ecosystem of TU Darmstadt, Hessian.AI and DFKI provides a unique environment to achieve this goal."
