10.1101/2022.07.18.500392

Local minimization of prediction errors drives learning of invariant object representations in a generative network model of visual perception