Some code and notebooks to help familiarise people with successor representation learning.
The SR_tutorial notebook uses a discrete state space and is the best place to get started.
The BVC_SR_tutorial notebook extends SR learning into using continuous features, and in particular focuses on using boundary vector cells as a basis representation as implemented in this paper "Neurobiological successor features for spatial navigation" https://www.biorxiv.org/content/10.1101/789412v2