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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Deep learning models are known to be overconfident in their predictions onout of distribution inputs. This is a challenge when a model is trained on aparticular input dataset, but receives out of sample data when deployed inpractice. Recently, there has been work on building classifiers that are robustto out of distribution samples by adding a regularization term that maximizesthe entropy of the classifier output on out of distribution data. However,given the challenge that it is not always possible to obtain out ofdistribution samples, the authors suggest a GAN based alternative that isindependent of specific knowledge of out of distribution samples. From thisexisting work, we also know that having access to the true out of sampledistribution for regularization works significantly better than using samplesfrom the GAN. In this paper, we make the following observation: in practice,the out of distribution samples are contained in the traffic that hits adeployed classifier. However, the traffic will also contain a unknownproportion of in-distribution samples. If the entropy over of all of thetraffic data were to be naively maximized, this will hurt the classifierperformance on in-distribution data. To effectively leverage this traffic data,we propose an adaptive regularization technique (based on the maximumpredictive probability score of a sample) which penalizes out of distributionsamples more heavily than in distribution samples in the incoming traffic. Thisensures that the overall performance of the classifier does not degrade onin-distribution data, while detection of out-of-distribution samples issignificantly improved by leveraging the unlabeled traffic data. We show theeffectiveness of our method via experiments on natural image datasets.

Improving robustness of classifiers by training against live traffic