Gradient descent aligns the layers of deep linear networks

Ziwei Ji, Matus Telgarsky. Gradient descent aligns the layers of deep linear networks. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net, 2019. [doi]

Authors

Ziwei Ji

This author has not been identified. Look up 'Ziwei Ji' in Google

Matus Telgarsky

This author has not been identified. Look up 'Matus Telgarsky' in Google