10.1101/2021.06.12.447639

Feedback alignment with weight normalization can provide a biologically plausible mechanism for learning