Interactive Learning of Grounded Concepts
This demonstration shows how an autonomous embodied agent can learn grounded concepts and words through task-based communicative interactions. The emergent concepts and words are fully transparent, generalize well, easily adapt to changes in the world, and can be learned in an incremental way.
Katrien Beuls received her PhD in Computer Science from the VUB in 2013. Within the VUB AI Lab, she is leading a research team on Evolutionary & Hybrid AI (EHAI - https://ehai.ai.vub.ac.be).