KI Data Tooling

Tutorial on Active Learning held by University of Kassel

What is Active Learning and how can it help the use of AI functions in autonomous vehicles? This was the topic of the tutorial on "Active Learning", organized by the University of Kassel.  The tutorial took place on March 18 via online event. In the course of the event, Daniel Kottke and Dr. Maarten Bieshaar from the department of Intelligent Embedded Systems explained in a first step what Active Learning is, what its objective is and what factors are necessary for Active Learning. Daniel Kottke focused on three selection strategies in particular.

He then explained that none of the aforementioned strategies could be used optimally so far, which is why the Intelligent Embedded Systems department is working on continuing und elaborating the existing models in their so-called Probalistic Active Learning method. Through this method, which he then presented in the tutorial, the team of the department hopes to eliminate some inaccuracies of the previously mentioned methods.

In conclusion, Daniel Kottke discussed current challenges and problems in the use of active learning and explained why active learning does not always achieve the resource-saving effect that the method actually promises. In addition, he pointed out a possible connection to Deep Learning, which has not yet been implemented in practice.

The numerous questions at the end of Daniel Kottke's presentation showed the hight interest in the topic and that Active Learning could have great potential for improving the implementation of AI functions. You can watch the entire tutorial on YouTube.