Learning from Demonstrations

The study ‘Preference-Conditioned Language-Guided Abstraction’ proposes a new way to improve learning from demonstrations in robots. It addresses the problem of spurious feature correlations that often occur in such learning…