Fetching the paper…

How Much Over-parameterization Is Sufficient to Learn Deep ReLU Networks? · Around