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