Dozent: Georgy Bagdasarov <lapskojs@gmail.com>

Dates: Blocks, To be announced!

Room: 111 Medienhaus

Credits: 2SWSt, 3 CP

The multiagency structure raises the question that intelligence emerges not only from the learning process of the entity itself, but rather through interaction between the vehicles of the entity and the environment. Through such separation—when an entity can distinguish its own species from the environment and intercommunication occurs between them—intelligence is created as a social rather than individual property.

Using deep learning methods, the plan for this course is to develop autonomous agents that can operate in a sonic environment. They will be driving and interacting with each other and with the environment, and through searching their paths, they will transform the environment itself. We will listen to their communication and maybe try to understand their ontology. For this, we don't need to know their language beyond Python, which will help us develop such systems that can work autonomously without human involvement. So at the very end, we can yell in excitement like a little kid who releases the bicycle handlebars—"Look Ma, no hands!"—right before he's about to crash.