Feedback Gradient Descent: Efficient and Stable Optimization with Orthogonality for DNNs

Open in new window