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 introduce a novel clean-label targeted poisoning attack on learningmechanisms. While classical poisoning attacks typically corrupt data viaaddition, modification and omission, our attack focuses on data omission only.Our attack misclassifies a single, targeted test sample of choice, withoutmanipulating that sample. We demonstrate the effectiveness of omission attacksagainst a large variety of learners including deep neural networks, SVM anddecision trees, using several datasets including MNIST, IMDB and CIFAR. Thefocus of our attack on data omission only is beneficial as well, as it issimpler to implement and analyze. We show that, with a low attack budget, ourattack's success rate is above 80%, and in some cases 100%, for white-boxlearning. It is systematically above the reference benchmark for black-boxlearning. For both white-box and black-box cases, changes in model accuracy arenegligible, regardless of the specific learner and dataset. We also provetheoretically in a simplified agnostic PAC learning framework that, subject todataset size and distribution, our omission attack succeeds with highprobability against any successful simplified agnostic PAC learner.

Broadly Applicable Targeted Data Sample Omission Attacks