Can overfitted deep neural networks in adversarial training generalize? – An approximation viewpoint

暂无描述。