Fetching the paper…

Learning without feedback: Fixed random learning signals allow for feedforward training of deep neural networks · Around