Learning without feedback: Direct random target projection as a feedback-alignment algorithm with layerwise feedforward training

Open in new window