Abstract Chrisantha Fernando 15th November 2012
Darwinian Neurodynamic Cognitive Architectures for Playful Robots
There is too much to do. On what basis should we decide what to do? Especially as neonates? There is too much to predict. On what basis should we decide what to predict? Especially as neonates. I want to make a robot that cries, and annoys its post-doc, until it is placed into an environment in which it can do and predict as it wishes. What should it wish? How should it wish? What is the algorithmic basis of enjoyment? What is a good predictor? Is it just reducing prediction error? No, can't be, because then predicting now from now would be the best and our brains would short circuit themselves in a useless and onerous self-obsession. How can brains reward a rich interaction and engagement in the world? The answer lies in the links between LEGO and KITTEN.
Back to IMC Seminar Series