How can we explain AI reasoning in images to humans?

Human-AI collaboration typically benefits from AI being able to explain its reasoning. While in computervision (CV) there are good methods to technically articulate the foundations of AI responses, they can hardly be understood by humans. Our approach is to link explanations to human-understandable concepts.

Now accepted at the CVPR (Conference on Computer Vision and Pattern Recognition) - one of the world’s most competitive tech conferences in the field. Noteworthy that the approach has been built on top of a master's thesis.

Congratulations to the authors Arne Grobrügge, Niklas Kühl, Gerhard Satzger, and Philipp Spitzer!

The full paper is accessible here:  Towards Human-Understandable Multi-Dimensional Concept Discovery