A renormalization group approach to (artificial) intelligence

2021. 10. 01. 10:15
online (Teams)
Antal Jakovác (Wigner)

The functioning of the renormalization group, in particular how it interpolates between different fixed points, is very similar to the understanding process of an intelligent actor (be it artificial or a natural one). Exploiting this analogy, we try to give a mathematical model of intelligence and understanding. We show how these thoughts work in scientific exploration, in image recognition, and we try to point out what is the main difference between the two. We also give some applications to demonstrate how these ideas can work in reality.