On the Implicit Bias of Dropout
proceedings.mlr.press › v80 › mianjy18bAlgorithmic approaches endow deep learning sys-tems with implicit bias that helps them general-ize even in over-parametrized settings. In this paper, we focus on understanding such a bias in-duced in learning through dropout, a popular tech-nique to avoid overfitting in deep learning. For single hidden-layer linear neural networks, we