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

Learning in adversarial settings is becoming an important task forapplication domains where attackers may inject malicious data into the trainingset to subvert normal operation of data-driven technologies. Feature selectionhas been widely used in machine learning for security applications to improvegeneralization and computational efficiency, although it is not clear whetherits use may be beneficial or even counterproductive when training data arepoisoned by intelligent attackers. In this work, we shed light on this issue byproviding a framework to investigate the robustness of popular featureselection methods, including LASSO, ridge regression and the elastic net. Ourresults on malware detection show that feature selection methods can besignificantly compromised under attack (we can reduce LASSO to almost randomchoices of feature sets by careful insertion of less than 5% poisoned trainingsamples), highlighting the need for specific countermeasures.

Is feature selection secure against training data poisoning?