Würzburg 2018 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: Arbeitskreis Physik, IT & KI (AKPIK) II

AKPIK 2.2: Hauptvortrag

Dienstag, 20. März 2018, 16:45–17:10, 70 - HS 00.107

Understanding the World with AI: Do we need a "CERN for AI"? — •Phillip Slusallek — Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI)

How can intelligent system support us in better understanding the complex world around us? And how can we use such systems to build better tools for science and many other applications?

Modern AI techniques like Deep Learning allow us to extract representations and models from big and complex data sets. If this data comes from sensors of the world around us, the resulting models form a sort of "Digital Reality" that allows us to simulate how we believe the world works. By simulating many relevant scenarios and through techniques like Reinforcement Learning then enables us to do learn strategies to reach goals in such environments. Finally, comparing the data derived from the simulations with data from the real world enables us to validate our models as well as incrementally improve them -- and thus our understanding of the world.

In my presentation I will discuss this Digital Reality in different domains ranging from (i)intelligent imaging sensors that incrementally adapt and refine their measurements based on prior knowledge and previous measurements, (ii) data analysis tools that use synthetic data from simulations to derive adaptive filters to scan large amount of sensor data for evidence, all the way to (iii) Autonomous Vehicles that need to understand the world around them to safely bring us to our destinations.