Training – introduction

The problem we need to solve is that we don’t know the alignment between states and observations.