English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We propose \emph{MaxUp}, an embarrassingly simple, highly effective techniquefor improving the generalization performance of machine learning models,especially deep neural networks. The idea is to generate a set of augmenteddata with some random perturbations or transforms and minimize the maximum, orworst case loss over the augmented data. By doing so, we implicitly introduce asmoothness or robustness regularization against the random perturbations, andhence improve the generation performance. For example, in the case of Gaussianperturbation,  \emph{MaxUp} is asymptotically equivalent to using the gradient norm of theloss as a penalty to encourage smoothness. We test \emph{MaxUp} on a range oftasks, including image classification, language modeling, and adversarialcertification, on which \emph{MaxUp} consistently outperforms the existing bestbaseline methods, without introducing substantial computational overhead. Inparticular, we improve ImageNet classification from the state-of-the-art top-1accuracy $85.5\%$ without extra data to $85.8\%$. Code will be released soon.

MaxUp: A Simple Way to Improve Generalization of Neural Network Training