Understanding deep learning requires rethinking generalization

  title={Understanding deep learning requires rethinking generalization},
  author={Chiyuan Zhang and Samy Bengio and Moritz Hardt and Benjamin Recht and Oriol Vinyals},
Despite their massivesize, successful deep artificial neural networkscan exhibit a remarkably small differencebetween training and test performance. Conventional wisdom attributessmall generalization error either to propertiesof themodel family, or to the regularization techniquesused during training. Through extensive systematic experiments, we show how these traditional approaches fail to explain why large neural networks generalize well in practice. Specifically, our experimentsestablish… CONTINUE READING
Highly Influential
This paper has highly influenced 66 other papers. REVIEW HIGHLY INFLUENTIAL CITATIONS
Highly Cited
This paper has 628 citations. REVIEW CITATIONS
Recent Discussions
This paper has been referenced on Twitter 5 times over the past 90 days. VIEW TWEETS

7 Figures & Tables



Citations per Year

629 Citations

Semantic Scholar estimates that this publication has 629 citations based on the available data.

See our FAQ for additional information.